Complementarity between private and public investment

in R&D: A Dynamic Panel Data analysis

Sadraoui Tarek, Naceur Ben Zina

To cite this version:

Sadraoui Tarek, Naceur Ben Zina. Complementarity between private and public investment in
R&D: A Dynamic Panel Data analysis. IAENG Conferences - WCE 2009 International Conference of Computational Statistics and Data Engineering, Jul 2009, London, United Kingdom.
<hal-00388525>

HAL Id: hal-00388525

https://hal.archives-ouvertes.fr/hal-00388525
Submitted on 26 May 2009

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.

Larchive ouverte pluridisciplinaire HAL, est

destinee au depot et `a la diffusion de documents
scientifiques de niveau recherche, publies ou non,
emanant des etablissements denseignement et de
recherche francais ou etrangers, des laboratoires
publics ou prives.

Complementarity between private and public investment

in R&D: A Dynamic Panel Data analysis
Tarek Sadraoui* & Naceur Ben Zina

Abstract This paper investigates the relationship between

private and public investment in R&D, while taking into
account the effect of several instruments policies such as
subsidies and taxes. We design a new look of knowledge
spillovers and R&D cooperation to explain the contribution of
public and private R&D on growth. We propose a
heterogeneous dynamic panel data model to consider the
endogenous effect of R&D investment. We also distinguish
between the estimated long and short run results. Our results
based on a sample of 23 countries over the period 1992-2004
indicate that both public and private investments in R&D are
complementary. By establishing an endogenous growth model,
the estimates indicate that public and private R&D depends on
the host countrys human capital investment. Results indicate
that foreign direct investment is a more significant spillover
channel than imports.
Key wordsR&D investment; Technology Spillovers;
Complementarities; Economic growth; Dynamic Panel Data;
Cointegration; Unit root test; Private investment; Public
investment; R&D cooperation

I. INTRODUCTION
Is public R&D complementary to private R&D, or does it
substitute for and tend to crowd out private R&D? Conflicting
answers are given to this question. We survey the body of available
econometric evidence accumulated over the past 16 years. A
framework for analysis of the problem is developed to help organize
and summarize the findings of econometric studies based on
dynamic panel data from various countries1. We conclude by
offering suggestions for improving future empirical research on this
issue.
Most people think that government R&D activities contribute
to innovation and productivity, many economists and policymakers
have grown frustrated with the paucity of systematic statistical
evidence documenting a direct contribution from public R&D (see
Paul et al,. [24]. Econometric findings concerning the productivity
growth effects of R&D seems to be that there is a significantly
positive and relatively high rate of return to R&D investments at
both the private and social levels (Cassiman and Veugelers, [8]). In
a recent survey, Paul et al,. [24] suggest that the especially
pronounced differential over the returns on tangible capital
investments observed at the private level may reflect individual
firms perceptions of especially high private risk in the case of R&D.
Public funding of R&D can contribute indirectly, by
complementing and hence stimulating private R&D expenditures.
Our approach will be to adopt a new econometric approach
using a dynamic panel data studies to analyse if public investment in
R&D are complement or substitute for private investment in R&D.
In literature review, we can conclude that the majority of the
econometric studies are concentrated on the impact of public R&D
contracts and grants upon private R&D investment by
manufacturing firms and industries (see for example, Lach [20],
Christopher [10] and Eric [12]).

The object of our paper is to give the theoretical and empirical

arguments which allow a satisfactory apprehension of the role that
the authorities must play in the fields of research and innovation.
The activity of R&D represents a significant source of development
of new knowledge and technological innovation (Guellec and
Van-Pottelsborghe, [14]). The effort towards activity of R&D
involves with a great importance and this through several resources
devoted to the various sectors and institutions of research.
Expenditure of research and development especially constitute a
principal source of growth of productivity for innovating countries.
Whereas, Sigrid [23] and Ting [31] suggest that, for countries,
where the activity of R&D misses almost technological knowledge
and innovations of which they profit are generally resulting from the
importation of equipment and goods of intensives investments in
technical progress. At the same times, Chaturvedi and Chataway [9]
recommend that knowledge capabilities and knowledge
management can be considered as key resources for firms in both
developed and developing countries.
II. R&D INVESTMENT AND THE GOVERNMENT'S R&D POLICIES
Today, we can observe an expansion of policies of innovations in
the developed countries which devote great investment for R&D.
What proves the creation of the climates favourable to the level of
these countries for the innovation? It is significant that during these
last years, companies of high technology or advanced technologys
(pharmaceutical, aeronautical) expenditure of research and
development increased significantly. The role of the governmental
policies as regards R&D is not to neglect. Indeed, the policies of
innovation define specific actions of the State, which must
encourage the accumulation of a qualified labour on the one hand,
and to help the companies to prospect the markets on the other hand.
This justifies the need for the public administrations for supporting
the R&D.
Thus, which are the reasons of the government aid and the
mechanisms the alternate ones available to the public
administrations to support the R&D? To answer these questions we
try to analyze the justification of the government aid with the R&D
starting from the economic theories of growth.
2.1

Neoclassic growth theory

For Neo-classic theory of growth, technical progress is

supposed to be exogenous factors. With the balance of long term,
population growth and technical progress determine the level of the
growth rate. This implies, according to the basic assumptions, that
the long-term growth rate is stable, and given in an exogenous way.
Within this framework, the impact of an action of the authorities is
practically ignored.
The neoclassic theory of the growth supposes that the economy
starts from a weak relationship between capital and labour. Just as
the marginal returns on capital are decreasing. What reduces the

encouragement to be invested in the new capital? Thus each new

unit of capital produces a lower income and less large savings. In the
long run, there will be absence of incentive to invest. In short, we
can say that the assumptions which underlie the neo-classic theory
are not realistic. The technological change is not always an
exogenous factor outside the market, determined by an unknown
process. To the 20th centuries, a good number of discoveries and
progress were carried out in the commercial sector by companies
with lucrative goal and not by public administrations or universities
where research is directed by non-commercial forces. Markets are
seldom in perfect competition, moreover, the private sector is not
capable to produce all the desired goods and services, because some
of them are goods public and certain others produce external effects.
2.2

Endogenous growth theory

The endogenous theory of growth recommends the relaxation

of certain neo-classic assumptions and incorporates the failures of
the market. However, the economic growth in the long run is
directed by the accumulation of the factors of production founded
this faith on knowledge, in particular, human capital, training, R&D
and innovation (Griliches, [13]). The endogenous models of growth
are characterized by a great diversity of the resources selected: The
investment in physical capital, in human capital, public capital, and
labour division, learning by doing, research and the technological
innovation (Romer, [27]).

2.3

R&D investment and market imperfection

Economic theory and empirical proof show that technical

progress, because of its incidence on the factors of production,
constitutes key element in the long run determining economic
growth; in certain countries, it represents even the most significant
element. However, it is not a question of an economic justification
of the official intervention for allocate the resources in favour of
R&D. But, this intervention in a market economy is justified by
incapacity of market to distribute resources in an efficient or
acceptable way as regards social aspects. With regard to the
investment in R&D, external effects and market imperfections
testify the incapacity of market, and the effects are felt not only
beyond particular companies but also beyond national borders.
In a market economy, a company will not invest in a project if
it knows that it can not adapt the possible receipts, however if it
cannot adapt a portion of these receipts, it will invest if this portion
is enough to make a profitable investment. Asymmetrical
information and imperfect competition constitute two other kinds of
imperfections of market involving under investment in R&D. For
example, asymmetrical information prevents effective operation of
capital market. Indeed, it can involve rationing of appropriations as
well as abandonment of investments in R&D in projects with strong
chances of success thanks to the plan of financing, and the
continuation of investments in the project having weak chances of
success.
III. RESEARCH METHODOLOGY
Our study contributes to the empirical literature -which is
discussed here- on the analysis of the existence of a relation between
private and public investment in R&D and their real effect on
economic growth; do public funds substitute or complement private
R&D expenditure?

We derive our econometric specification from a function

including interactions between internal and external R&D in the
augmentation of the knowledge stock. The model also takes into
account potential productivity convergence by including lagged
productivity levels. Our studys inferences are based on a dynamic
panel data model, which allows us to control for the existence of
unobserved fixed effects that are likely to affect R&D decisions.
Estimation is carried out by several consistent dynamic panel data
methods, among which generalized method of moments, which
allows for the presence of weakly endogenous explanatory
variables. In this way the analysis can take into account both degree
and possibility effects of R&D to address the issue of optimal
combinations of R&D expenditures.

In this paper, we contribute the first panel data study

exploring complementarity between public R&D and private
R&D in a dynamic panel framework. We examine the impact
of internal and external R&D on economic growth in
sixteen-year panel for 23 developed and developing
countries.
IV. COMPLEMENTARITY VERSUS SUBSTITUTABILITY BETWEEN
PRIVATE AND PUBLIC R&D
Theoretical work did not succeed in slicing on favourable or
unfavourable effect using certain political instruments on the level
of R&D in private sectors. The results of each model strongly
depend on its structure and its assumptions. Empirical work, leads to
homogeneous results and identifies a positive effect of public R&D
on that private (Paul, et al,. [24]). With an aim of knowing the
relation between public and private R&D we give an overall picture
of the activities of R&D in world. Indeed, in this section, we attach
more importance to activity of public and private R&D in the most
significant poles in world.
Through time and with the improved scientific methods in
particularly studies of Jason, et al., [19], it became clearly that the
final situation towards the effect of the public funds of R&D cannot
be made. Thus, in general, two fields can be identified and which are
used to analyze the relation between private investment and public
in research and development with knowing quantitative and
qualitative studies: On the one hand; for the qualitative studies, data
are frequently based on the investigations. On the other hand; for the
quantitative studies, they are based on macro and micro-economic
information of a significant number of companies (Cassiman and
Veugelers, [8]).
Today, several activities of R&D are carried out on the level of
the services sector. On the one hand, this is due to the external
sources of the strategies of manufacturing industries in the Eighties.
On the other hand, the transformation of information and
technology of communication get more opportunities for innovating
sectors. So the governments help more and more activities of R&D
in several sectors with an aim of stimulating technological
performances of their countries. Thus, several examples can be
quoted. At this level, for the Nineties and more precisely in 1999,
the total expenditure of R&D of Germany is 47 billion dollars where
66% of this amount is invested by private industries, 18% by
government and the remainder are invested by foreign companies.
Thus, Claudia, [11] suggest that an international comparison
on behalf of public programs of R&D shows that Germany is one of
principal countries which grant funds for the technological
performance. At this level, manufacturing industry plays a very
significant role concerning R&D. For example, the strategic
planning of the national research evaluation in Thailand as indicates

it Jarunee, [18] is to allocate the budget to support the research

programmes and projects. Jarunee suggest that to improve the model
evaluation framework for R&D investments, the public hearing
forum was organised. From there, a question emerges up to what
point evolution of public funds of R&D makes it possible to
stimulate R&D carried out by private sector, and on which level
results are checked? Recently, an econometric micro study tackled
the question of the impact of political instruments about activity of
R&D deprived on the level of companies.
Several other studies are more precisely interested in testing
the effects of public subsidies in R&D on the amount of deprived
investment see for example Lai et al., [21], Hans and Almas, [16]
and Christopher, [10]. The major goal of these studies is to know if
public subsidies of research and development can have an effect of
reduction or increase in the expenditure of R&D. Most of results
suggest that public subsidies of R&D on the level of several
industries showed that there is a small tendency to the effect of
ousting Crowding out. In addition, it seems not to have any effect
or degree of Complementarity.

In the following section we empirically test fundamental

relation which we seek to analyze in the case of 23 countries
for the period 1992-2004, in other words we test the
existence of a relation of Complementarity and to check this
result.
V. EMPIRICAL VALIDATION: DYNAMIC PANEL DATA
The objective of this work is to test the impact of an action of
public policies empirically on the evolution of R&D in private
sector while trying to surmount limits. The modelling that we follow
to measure the effect of the R&D deprived on the public one; while
taking into account some determinants of private R&D; is the one of
Bettina and al,. (2002). This modelling has also been applied by:
Busom, [7] and Lach, [20]. The gait of these authors can be
summarized like follows:

research and development of country (i) are explained by stocks of

research of the period (t-1).
X: Represent the vector of exogenous variables; these variables are
added value (VA), public research (G), import (M), foreign direct
investment (IDE) and private research;
(, ): Designate parameters to estimate;
i : Constitute individual heterogeneity as: i i.i.d. ~ N [(0, 1)];
And: vi,t is stochastic term as: vi,t~ i.i.d. [(0, 1)].
yi,t is the logarithm of volume of R&D in country (i).
xi,t is determinant vector of R&D.
i is the specific effect of country (i). This specific effect can be a
stationary or uncertain effect.
5.2 Evaluation Method
The evaluation of the model by traditional methods (Ordinary
Least Square "OLS" and within) gives biased and non convergent
values because of inter-relationship between retarded endogenous
variable and individual heterogeneity. We try to demonstrate for the
case of a simple model the inconveniences of these methods of
evaluations.
For dynamic panel model, Within transformations and Ordinary
Least Squares are biased and non-convergent estimators.
We assume the simple specification:

y it y it -1 i it

Let us pose

y y

a =

The underlying logic is simple: If the coefficient * has a

positive sign we can say that public R&D are complementary for
private R&D. In other words, an increase of 1% of public research
and development level entails a growth of *% of private R&D. On
contrary, if * has a negative sign we can say that there is a relation
of substitutability between public and private R&D. In this part we
try, to give a general setting for the models to estimate while putting
accent on some remarks and inconveniences of these models. We
apply a dynamic panel data model. Finally, after having estimated
the model we analyze results.

5.1 Dynamic panel data: Definition and evaluations

method
Dynamic models are characterized by presence of one or
several endogenous variables delayed among explanatory variables.
Our specified model is a dynamic panel model is given by:

y it y i t -1 ' x i t v i t

(1)

Under another forms one was writing our model as below:

p lim
N

y it -1 : Endogenous variable appears in the regression as being a

retarded explanatory variable. In other words, present stocks of

i= 1

y2
t= 1

it - 1

i 1

t 1

i 1

it - 1

1
NT

i 1

t 1

i v it y
N

i 1

2T

cov

p lim
N

t 1

it 1

it 1

t 1

it 1

y , T (1 ) (T 1) T

i0

T
1
iN 1 y 2
it 1
NT
t 1

N
2 1
1 1 2t i 1 yi0
1 2t 1 2t

T 2

1 1 2
T 1
N
T (1 )2 T

it 1

y2
N

i v it y

1
NT

= 1 1
T 1

it

t= 1
T

i v it y

is convergent if

p lim

R it R it -1 1G it 2 M it 3VAit 4 IDE it i it (2)

Where;

i= 1

is stationary then:

y it

Private R&D = * public R&D + control variables + e

and that

1 2t 1 2t
2

cov( y i0, i)
T (1 ) 1
1 2

( T 1) T y 2 2 T

T ((1 ) 2 ) 2

In summary, the bias is positive and increases with the variance

of the specific effect. Indeed, yi,t is function of vi,t and yi,t-1 is also.
yi,t-1 is an explanatory variable correlated with stochastic term. It
introduces a bias in the value of ordinary least squares. Even as
putting hypothesis that stochastic terms are not correlated, this value
is non-convergent.
For within case we consider the following transformation:

y it - y i . y it - 1 - y i . - 1 it - i . - 1
With:

E v v
it js 0 s i i j e t s i t s

ifn o t
E v it v js 0
While posing as in Baltagi [3]:

y it 1
T 1

y i. 1

T
iN 1 yit y i. y it 1 y i. 1
t 1

T
2
tN 1 y it 1 y i. 1
i 1

p lim
N

s
T

(T

y it 1 y i . 1

i 1

t 1

1) -

T a

(1 -

a )

1
p lim
N NT
=

1-

v
2

i 1

1
T

t 1

it 1

it

i.

Where

(5 )

We consider the case where temporal dimension is small while

individual dimension (N) is important. We consider that individual
effects are stationary and we assume traditional hypotheses of
residues:
In difference models (5) can be written as below:

yit yit -1 uit

E yit-yit-1 yit- j 0

i .1

Anderson and Hsiao [1] proposed, initially, to write the model

from first difference to eliminate individual heterogeneity. They
propose for the transformation two instruments.
T
iN1 yit - yit -1 yit -2- yi.-3
t 3
vi (1)
T
N
i1 yit - yi.-2 yit -2- yi.-3
t 1

And

(4)

(6)

Where uit = vit-vit-1.

5.3 Anderson and Hsiao Method

The two values are convergent when N and T .

However, an inter-relationship always persists between endogenous
variable in first difference and residual term. Authors proposed to
resort to the method of instrumental variables to surmount this
problem. Thus, they propose to use instrument endogenous variable
with two lags or his first differences. These instruments are
correlated with explanatory variable and are not with residual term.
To get more efficient results, Arellano and Bond [2] approach
permits to get a value of generalized moments GMM more
efficient.
5.4 Arellano and Bond Approach

The estimator of Least Square Dummy variable is convergent

if T is infinite. If T is fixed, N, the estimator is
non-convergent. Our model should not be estimated by the
method of OLS and LSDV due to the fact that estimating by
these methods led to ad hoc results. Which are then adequate
methods to estimate our model? We propose below two
methods which consist in obtaining consistent estimators.

We test for every individual of the linear restrictions of type:

(T - 1) - T a +

2a

(1 - a )

And the denominator

y it y it -1 i v it

T y it 1 y i. 1 it i.
iN 1
NT
t 1

2
T y
y
iN 1 it 1 i. 1 NT
t 1
The numerator is convergent when the second term converges
towards zero.
The numerator of the second term:
1
NT

Arellano and Bond [2] are the first in 1991 that proposed an
extension of GMM introduced initially by Hansen [15], to the case
of panel data for a simple model AR (1):

Thus we can write that:

T
iN 1 y it - y it -1 y it - 2
t3
v i ( 2 )
T
N
i 1 y it - y i .- 2 y it - 2
t 1

(3)

for j 2,..., t; t 3,...T (7)

The gait of Arellano and Bond, in presence of the exogenous

variables, consists in estimating the model in difference:
p
(8)
y it k y i ( t - k ) '( L ) X it v it
k 1
Moment conditions and instruments matrix are given respectively
by:

E y
0
it v is

E X it v it 0
0
0
Z ip

Z ip1 .
Z
.
.
.

X ip 2 X ip3 .

pour 2, t 2, 3, ....T

(9)

pour 2, t 2, 3, ....T

... . . X iT

.. . .
.. . .
.. . .

0
.
.

(10)

The preceding dynamic model (8) can be rewritten for each

individual in the following form:
y i W i i i V i
(11)
Where is a vector of parameter and Wi is a matrix that
contains the retarded dependent variable and explanatory variables.
The method proposed by these authors permits to get a GMM in
two stages is written in following form:

-1

*
*'

W
Wi Z i AN Z 'i i W*'
AN Z iy*i
Z
i
i

i
i
i
i

(12)

However, to have previous value GMM, it is necessary to pass

by a first stage that consists in making wished transformation (first
difference or orthogonal deviation), to find and to use instruments
matrix and to achieve a first evaluation named "evaluation of first
stage". This stage corresponds to an evaluation that permits to
provide estimated residues after transformation. In the first stage,
the values are gotten while using Hi as:
H i v *i v *i '
(13)
2 -1 ............. 0

-1 2
.
.
Hi

.
-1
0 .
0 0 ............-1 2

-1

Z ' I N H Z

y i1
0
.

x i1
0
.

xi 2
0
.

y i 1 x i1
. .
0 0

xi 2
.
0

xi3
.
0

..
..

0
0

..

..

y i1

y i ( T -2 )

0 .... 0
0 ..... 0
. ..... .
x i1 .... ...

x i ( T -1)
0
.
.

Arellano and Bond [2] propose a test verifying the absence of

autocorrelation of first and second order. Thus, if distribution is non
auto-correlated, this test gives a value of residues differentiated
negative and significant to first order and non significant to the
second order. This test that is based on auto-covariance of residues
follows a normal law N (0,1) under hypothesis H0. Otherwise,
authors propose the test of validity of instruments of Sargan
(1988a). The statistical test equal to:
S

v *i ' Z i A N Z ' i v *i

(1 5 )

VI. EMPIRICAL RESULTS

.
(17)

1 N
t i
N i1
Where ti is the individual t-statistic of testing Ho: i = 1 in (18). It
is known for a fixed N as T
t

ti

W dW

iz

iz

W2
0 iz
1

1/ 2

tiT

(18)

IPS assumes that tiT are iid4 are have finite mean variance. Then;
1 N

N
tiT E [ tiT / i 1
N i 1

N (0, 1)
Var [ tiT / i 1 ]

(19)

As N central limit theorem. Hence

t IPS

N t E [ tiT / i 1
N (0, 1)
Var [ tiT / i 1 ]

(20)

As T followed by N sequentially, the values of

E[tiT/i=1] and Var[tiT/i=1] have been computed by IPS
simulations for different values of T and is. As applying test on
our complete model our results is summarized in table 1
VII. CONCLUSION

The unit root tests became a current step for analysis of time
series stationnarity. However, practical application of these tests on
panel data is recent. The tests most frequently used are those of
Levin and Lin, [22] and of Im, Pesaran and Shin [17]3. Recently,
several procedures of unit root tests and Cointegration were
developed for panel data models. The addition of individual
dimension to temporal dimension offers an advantage, in practical
application of unit root and Cointegration tests (Pedroni, [25, 26]).
The checking of non-stationary properties for all panel
variables leads us to study the existence of a long run relation
between these variables. The Cointegration study by applying
Pedroni Cointegration tests based on unit root tests on residues
estimated. Cointegration tests on panel data consist in testing the
presence of unit root in the estimated residues. However, the
problem of fallacious regressions, of the time series, also arises in
the case of panel data. First step is to test unit root for each of series.
6.1 Unit Root Tests
Levin and Lin [22], consider the following model:
yi,t = iyi,t1 + Zit + ui,t (i=1, , N; t=1, , T)

ij u it j it

The null and for all countries i the alternative hypothesis are:
Ho: i = 1
Ha: i < 1
For at least one i, the IPS t-bar statistic is defined as the average of
the individual ADF statistic as:

(14 )

0
0

pi
j 1

pi

y i , t i y i t 1 j 1 ij y it j z it
it

The objective of transformation is, as at Anderson and Hsiao

[1], to eliminate individual heterogeneity of the model. The number
of instrument increases in the time for every individual. In the case
where exist explanatory variables xit in the model correlated with
heterogeneity individual i. Optimal instruments matrix
corresponding Zi is equal to:

u i ,t

Substituting this ui,t in (1) we get:

And

1
A N Z i' H i Z i
N

Where, Zi,t is the deterministic component and ui,t is a stationary

process. i is the fixed effect,
The Levin and Lin, [22] tests assume that ui,t are iid (0,2u) and
i= for all i. The LL test is restrictive in the sense that it requires
to be homogeneous across i. Im, Pesaran and Shin [17] (IPS) allow
for a heterogeneous coefficient of yi,t-1 and propose an alternative
testing procedure based on averaging individual unit root test
statistics. IPS suggested an average of the augmented Dickey-Fuller
(ADF) tests when ui,t is serially correlated with
different series. Correlation properties across cross-sectional units,
i.e;

(16)

In our survey, we tried to put accent on private and public

investment in R&D, for the case of 23 countries which presents
different levels of R&D. We tried to clarify relation that exists
between private and public research. This empirical survey wanted
to give account, the effects of different determinants on private
investment in R&D and to know if public and private investments in
R&D are complement or substitute.
Econometric approach consists in the regression of some
measures of private R&D on public R&D with some control
variables. The evaluation that we presented in our work corresponds
to GMM evaluation in first difference and in orthogonal deviation.
We prefer to refer to results of this evaluation because it permits to
eliminate rigorous way all bias to none observed individual
heterogeneity and offer, a better efficiency of results. Empiric
evaluations confirm a positive effect of public R&D in different
country (positive and meaningful effect in all evaluations). Results
of our empiric survey are relative for our sample and they go in the
sense of results of ulterior studies, which showed that there is a
positive and meaningful relation between private and public
investment in R&D.

All results are in favor of a positive relation between private

and public R&D which can be assumed by a complementarities
between them. In our study we have indicate that all variables are
stationary by an application of unit root tests thats can contribute to
search Cointegration relation between them and determinate the
number of these relation. Another important think, we can give the
impact of public R&D to private R&D for each country to specify
the nature of relation and how private R&D contributes for public
sector. In summary, all countries must stimulate private sector in
R&D activities to promote economic growth and integrate a new
innovation system which can go with their own economic
environment.
Some studies put in value of other factors that can be important
as: competition in the market, public politics and cooperation
concerning R&D between firms. Cooperation in R&D is a part of
the new strategies developed by firms in more global and
competitive economic environment. These last factors are not to
disregard and can be subject of a future research concerning the
relation between public and private investment in R&D.
APPENDIX

Table 1Unit root Tests

Statistics

VA

IDE

Levin-Lin
ADF stat
IPS ADF stat

-2.357

-0.343

1. 809

1.489

-1.674

-2.296

1.773

2.175

1.640

-1.572

REFERENCES
[1] Anderson, T.W. and Hsiao, C. (1982) Formulation and Estimation of
Dynamic Models Using Panel Data, Journal of Econometrics, Vol. 18, pp.
47-82.
[2] Arellano, M. and Bond, S. (1991) Some test of specification for panel
data: Monte Carlo evidence and application to employment equations,
Review of Economic Studies, Vol. 58, pp. 277297.
[3] Baltagi, B.H., James, M.G. (1995) A Dynamic Demand Model for
Liquor: The Case for Pooling, The Review of Economics and Statistics, Vol.
77, N. 3, pp. 545554.
[4] Becker, B. and Pain, N. (2002) What determines industrial R&D
expenditure in the UK?, National Institute of Economic and social research
NIESR, Discussion Paper No. 211, April.
[5] Blank, D. and Stigler, G. (1957) The demand and supply of scientific
personnel, New York, National Bureau of Economic Research.
[6] Bernard, C.B. and Guadalupe, S.D. (2007) Innovation and R&D
spillover effects in Spanish regions: A spatial approach, Research Policy,
Vol. 36, pp. 13571371.
[7] Busom, I. (2000), An empirical evaluation of the effects of R&D
subsidies, Economics of Innovation and New Technology, Vol. 9, pp.
111148.
[8] Cassiman, B. and Veugelers, R. (2006) In search of complementarity in
Innovation Strategy: Internal R&D, Cooperation in R&D and external
technology Acquisition, Management Science, Vol. 52, No. 1, pp. 6882.
[9] Chaturvedi, K. and Chataway, J. (2006) Strategic integration of
knowledge in Indian pharmaceutical firms: creating competencies for
innovation, Int. J. Business Innovation and Research, Vol. 1, Nos. 1/2,
pp.2750.
[10] Christopher, A. L. (2005) Market structure and endogenous
productivity growth: how do R&D subsidies affect market structure?,
Journal of Economic Dynamics & Control, Vol. 29, pp. 187223.
[11] Claudia, S. (2008) Complementarities of innovation activities: An
empirical analysis of the German manufacturing sector, Research Policy,
Vol. 37, No. 9, pp. 14921503.
[12] Eric, C.W. (2007) R&D efficiency and economic performance: A
cross-country analysis using the stochastic frontier approach, Journal of
Policy Modeling, Vol. 29, No. 2, pp. 345360.
[13] Griliches, Z. (1995) R&D and Productivity: Econometric Results and
Measurement Issues, in Paul Stoneman (ed.), Handbook of the Economics of
Innovation and Technological Change, Blackwell, pp. 52-89.
[14] Guellec, D. and Van-Pottelsberghe, B. (1997) Does Government
Support Stimulate Private R&D?, OECD Economic Studies, Vol. 29 No. II,
pp. 95122.

[15] Hansen, L.P. (1982) Large Sample Properties of Generalized Method

of Moments Estimators, Econometrica, Vol. 50, N. 4, pp. 1029-1054.
[16] Hans, L. and Almas, H. (2005) The impact of public funds on private
R&D investment: New evidence from a firm level innovation study MIT
Discussion Paper, ISSN 1795-5300, N. 3, pp. 126.
[17] Im, K., Pesaran. H., Shin, Y. (2003) Testing for unit roots in
heterogeneous panels , Journal of Econometrics, Vol. 115, pp. 5374
[18] Jarunee, W. (2008) Building effective research evaluation systems to
assist R&D investment decisions, International Journal of Business
Innovation and Research, Vol. 2, No. 2 pp. 123 140.
[19] Jason, C.H. and Eduardo, S.S. (2008) A model of R&D valuation and
the design of research incentives insurance, Mathematics and Economics,
Vol. 43, No. 3, pp. 350367.
[20] Lach, S. (2000) Do R&D subsidies stimulate or displace private R&D?
Evidence from Israel, NBER working paper n7943, Cambridge.
[21] Lai, M., Peng, S., Bao, Q. (2006) Technology spillovers, absorptive
capacity and economic growth, China Economic Review, Vol. 17, pp.
300320.
[22] Levin, A. and C.F. Lin, (1992) Unit root tests in panel data: asymptotic
and finite sample properties, Department of Economics, University of
California at San Diego, Discussion paper , Nos. 92-93.
[23] Lopez-Bazo,, E., Requena, F., Serrano, G. (2006) Complementarity
between local knowledge and internationalisation in regional technological
progress, Journal of Regional Science, Vol. 46, pp. 901930.
[24] Paul, A.D. Bronwyn, H. H., Andrew, A. T. (2000) Is public R&D a
complement or substitute for private R&D? A review of the econometric
evidence, Research Policy, Vol. 29, pp. 497529.
[25] Pedroni, P. (1999) Critical Values for Cointegration Tests in
Heterogeneous Panels with Multiple Regressors, Oxford Bulletin of
Economics and Statistics, Vol. 61, pp. 653 670.
[26] Pedroni, P. (2004) Panel Cointegration: Asymptotic and Finite Sample
Properties of Pooled Time Series Tests with an Application to the PPP
Hypothesis, Econometric Theory, Vol. 3, pp. 579 625.
[27] Romer, P. (1990) Endogenous technological change, Journal of
Political Economy, Vol. 98, No. 2, pp. 71102.
[28] Sargan, J. D. (1988a) Contributions to Econometrics, Vols. 1 & 2. (Ed.
E. Maasoumi). Cambridge: Cambridge University Press.
[29] Sigrid. S. (2005) Cooperative and noncooperative R&D in
experimental duopoly markets International, Journal of Industrial
Organization, Vol. 23, No.12, pp. 6382.
[30] Tahir, C., Ozalp, V., Bahar, S. (2008) A combined decision model for
R&D project Portfolio selection, International Journal of Business
Innovation and Research, Vol. 2, No. 2 pp. 190 202.
[31] Ting, G. (2005) Foreign direct investment and growth under economic
integration, Journal of International Economics, Vol. 67, No 1, pp.
157174.
[32] Veugelers, R., Cassiman, B. (2005) R&D Cooperation between Firms
and Universities, some empirical evidence from Belgium Manufacturing,
International Journal of Industrial Organisation, Vol. 23, No. 56, pp.
355379.
[33] Wieser, R. (2005) Research and Development Productivity and
Spillovers: Empirical Evidence at the Firm Level, Journal of Economic
Surveys, Vol. 19, pp. 587621.
[34] Xulia, G. and Consuelo, P. (2008) Do public subsidies stimulate private
R&D spending?, Research Policy, No. 37, pp. 371 389.