ISSN: 2158-7051 ==================== INTERNATIONAL JOURNAL OF RUSSIAN STUDIES ==================== ISSUE NO. 3 ( 2014/2 ) |
COMMENTS ON THE IMPACT OF KNOWLEDGE ON ECONOMIC GROWTH ACROSS THE REGIONS OF THE RUSSIAN FEDERATION
JENS K. PERRET*
Summary
Using a basic growth accounting approach it is
deduced how far the regional knowledge infrastructure plays any significant
role across the regions of the Russian Federation. Aside from aspects of the
size of the regional innovation system, like the number of researchers and
students, it is discussed in how far the inflow and outflow of knowledge plays
a role in determining the economic growth. The study shows thereby that while the Russian
growth dynamics are indeed driven by the exploitation of natural resources,
foremost of oil and gas, a significant part of Russian growth is due to its
innovation system. This shows that innovation oriented growth politics as
promoted by former president Dmitry Medvedev do have a solid foundation to be
built on.
Keywords: Economic Growth, Russian Federation,
Knowledge, Innovations.
Introduction
Not only since the works by Machlup[1],
who described the modern society as a knowledge society, has the aspect of
knowledge as a production factor and an essential building block of economic
growth been acknowledged.
A broad range of studies exist that have reported empirical results on
the importance of knowledge in economic growth. However most
of these studies focus on Western Europe, the European Union, its member
states, the USA or other highly developed economies. A smaller range of
studies focuses on how knowledge, and in specific the inflow of knowledge, can
facilitate economic growth in developing economies, especially those in
transition.
Internationally the Russian Federation is generally considered as a
provider of basic resources or low quality goods. Only a few authors have
considered the Russian innovation system and thereby the contribution of
knowledge to Russian economic growth. In some part this shortcoming is
motivated by the lack of suitable data.
In the present study this research gap is filled by providing an insight
into the effects regionally domestic knowledge sources, as well as
intra-regional knowledge flows, have on Russian
economic growth.
In the following second section the research design is presented while
in the third section the results from a dynamic spatial panel regression are
presented and discussed before in the fourth section some preliminary
conclusions are drawn.
Knowledge Extended Growth Accounting
Using a growth model approach, this section argues how different sources
of knowledge, knowledge spillovers and the absorptive capacity, influence the
economic situation in the Russian Federation, measured by the GRP. A number of
studies like Guellec and and van Pottelsberghe de la Potterie (2004) exist that
analyze the importance of R&D and the institutional environment on the
output of an economy. For the Russian Federation, however, Ahrend (2002) argues
that political and institutional features are almost unimportant and can
therefore be left out of a growth-related analysis.
In addition to the knowledge inputs generated inside the region,
knowledge inputs generated outside the region that enter the region in the form
of interregional spillovers and through international channels of knowledge
transfer like foreign trade or direct investments are considered.
From a theoretical point of view the model picks up on the Solow growth
model, assuming an influence of the labor and capital inputs on the level of
the GRP. However, the most interesting aspect lies in modeling the Solow
residual.
Finally, the new economic geography stresses the importance of the
underlying spatial structure, motivating thereby the implementation of spatial
models.
The model underlying the following estimations is always considered to
be in log-linear form. Therefore, all variables, as long as they do not
represent a quota or percentage, are logarithmized versions.
As the state-owned sector makes up a comparatively large share of total
production, but might be considered less efficient than the private sector[2],
and the amount of government personnel can also be interpreted as a proxy
variable for corruption, which should also have a negative effect on growth,
government personnel is included as a control variable[3].
Partially hand in hand with the importance of state-owned firms goes the
share of natural resources in the Russian economy, of which the oil and gas
sector comprises a significant share; therefore, the amount of produced oil and
gas is included into the model as well. Here the hypothesis of the resource curse can be recalled, as it
proclaims a negative relation between the development of the GDP and the amount
of non-renewable natural resources - especially natural oil and gas[4].
The knowledge input side of the economy is represented by four
indicators: the number of researchers, the expenditures on R&D, the number
of students and the number of patents granted. As spillover effects are to be
included in the model as well, but respective indicators are only available for
patents granted by the EPO, only patents granted by the EPO are considered in
the analysis.
Finally, from an international perspective, the exports and imports as
well as the openness indicator - to give a general insight into the integration
of a region into the world economy - are considered[5].
However Lichtenberg and and van Pottelsberghe de la Potterie (1998) argue that
it is not so much the intensity of the imports, and thereby the exports as
well, but the distribution of the countries of origin or their destination that
influence economic development. The trade related indicators are accompanied by
the amount of FDI inflows - as another channel through which knowledge can
enter a region[6].
Running a series of tests on a first basic model reveals that only the
fixed effects model will produce reliable estimates for the model, and it also
suffers from heteroskedastic error terms. The ongoing analyses therefore rely
on robust standard errors.
Application of a Moran's I test and robust Lagrange multiplier tests for
spatial autocorrelation effects reveal significant spatial autocorrelation
effects.
Further testing shows that the model suffers from serial autocorrelation
as well. Summarizing these results leads to the use of the
Blundell-Bond estimator for dynamic panel models in the context of a Han
Philips Spatial Dynamic estimation method[7].
To account for a structural break, which is rather likely in the event
of the crisis in 1998, the total time frame has been divided into the
transition years including 1998 and the later years starting with 1999. To test
for a structural break in levels a dummy variable for the transition years is
included in a first regression (model I). In two other regressions the
transition (model II) and the later years (model III) are considered
separately.
Since the correlation matrix for the independent variables suggests that
some of the variables are highly correlated, variance inflation factors are
calculated, revealing that severe problems with multi-collinearity exist.
Testing different reduced versions of the basic model leads to the result that
it can be cleaned of multi-collinearity - or rather of variables reporting variance
inflation factors larger than ten - by omitting the labor and capital
variables, which are highly correlated with each other as well as with the
researchers, R&D expenditures and government personnel.
The expenditures on R&D have been removed as well since they are
highly correlated with the researchers and the government personnel.
As a fourth variable, either the researcher or the government personnel
variable needs to be removed from the equation. While removing the researchers leads to a qualitatively better model in
general, their removal would also exclude an essential insight on the influence
of the tacit knowledge potential on the economic development across regions.
Therefore, two basic models have been estimated - one with researchers and one
with government personnel. The model implementing government personnel is
considered as well as a stability test for the results of the researcher model.
While it can be argued that the approach is no longer valid since labor
and capital variables as base variables of the underlying production function
structure had to be excluded, the approach here can be seen as measuring the
effect of mostly knowledge-oriented inputs that influence economic growth aside
from labor and capital, which are natural drivers of economic development and
growth nonetheless. Referring to the neoclassical growth model, this reduced
version is basically an approach to quantify the Solow residual.
Empirical Analysis
While the patent variable and the spillover variables are based on
patent data by the European Patent Office all other variables are based on the
regional statistical yearbooks by Rosstat.
Data has been used for the years 1994 to 2009 in a first model which
does not include spillover effects and for the years 1994 to 2006 in a second
model which includes spillover effects. Therefore, the estimation considers
sixteen or thirteen years and 80 cross-sections each[8]
leading to a total of 1,280 or 1,040 observations respectively.
Table 1: Regression Results
using Researchers
Table 2: Regression Results
using Government Personnel
For reasons of multicollinearity researchers and government personnel
are not implemented together, therefore Table 1 summarizes the results for the
model without spillover effects including the researcher variable while Table 2
summarizes the results for the model without spillover effects including the
government personnel variable[9].
Comparing the tables there is no big difference in the signs of the
coefficients whether researchers or government personnel are used as a
variable. Quality indicators like the R2 and the F-test yield
similar results for the lag-model; with the Durbin model the F-statistics are
significantly larger in the case of using the government personnel variable.
However, they are a first indicator that the results are stable.
Additionally, when comparing the results for the spatial lag and the
Durbin model the variables retain their signs even though a few lose their
significance the quality statistics indicate largely comparable results.
Considering the signs of the variables themselves most of them represent
results expected from economic theory. The only exceptions are the positive
impact of imports, the negative impact of openness and the positive impact of
the government personnel.
However, the positive impact of imports can be explained by assuming
that the positive relation is not a direct effect but rather represents the
effects of an antecedent variable. Having a better economic situation in a
regions leads on the one hand to higher GRP values and on the other it leads to
a larger number of wealthy inhabitants who in turn are more interested in
acquiring foreign products.
On the other hand it can be assumed that a region that is producing
efficiently also requires quality equipment which in turn is imported from
abroad - a similar argument holds for foreign direct investments as well which
on the one hand raise the economic output of a region but on the other hand
might lead to higher imports of intermediate goods.
Nevertheless, a peculiar result is the consistently positive impact of
the government personnel which, from the perspective that state ownership
generates less efficiency as well as from the perspective that the number of
government personnel can be used as a proxy for the level of corruption, can be
seen as counter intuitive. Especially, since the comparatively large
coefficient implies that a doubling of the amount of government personnel will
lead to unrealistically high growth rates. This effect might be generated via
the large share of government activity in the sector of natural resources which
biases the analysis from the start.
Considering that the spatial lag variable is highly significant in the
spatial lag model and that most of the spatially lagged variables in the Durbin
model are highly significant this shows that there are important links between
the regions. This goes along with the consistently positive and highly
significant lagged variable which shows that economic growth across the regions
of the Russian Federation is highly path-dependent.
Regarding the size of the coefficients the most important result of this
study stems from comparing the researchers and the oil and gas production
coefficients. While a doubling of the amount of produced oil and gas only
results in raising the GRP by roughly 5% to 7%, a doubling of the amount of researchers results in raising the GRP by roughly 25% to 32%[10].
Even considering that the results might be biased by measurement errors it
shows the remarkable importance of the research sector for the economic development
of the Russian regions. Furthermore, it strengthens the hypothesis that
investments in Russian high technology, research intensive sectors is not only
more sustainable in the long term, but has consistently - even in the
transition years - been driving the Russian regional development process.
Additionally, the importance of FDI inflows is even less significant for
Russian growth as a rise of FDI inflows by one percent only leads an additional
0.01% of economic growth. However it needs to be considered that on average it
is much easier to double the inflow of FDI than doubling the output of the oil
and gas sector.
Table 3: Regression Results
using Researchers – Extended Model
Table 4: Regression Results
using Government Personnel – Extended Model
The consistently significant spatial effects reported in Tables 1 and 2
show that there needs to be some kind of interregional link between the
regional entities of the Russian Federation. Assuming this link to be based on
the diffusion of knowledge is, at least from a Western European perspective, a
viable option.
If the time horizon is cropped to allow for the use of the patents at
the EPO[11]
and active as well as passive patent citations[12] -
as proxies for the in- and outflow of codified knowledge - and inventor inflows
and outflows[13]
- as proxies for the in- and outflow of tacit knowledge[14].
Tables 3 and 4 capture the results of the extended models[15].
The basic impacts of the variables that were previously implemented do
not change and the qualitative indicators also do not change significantly.
Thus, knowledge spillovers are rather unimportant for Russian regional growth
and Russian regional growth benefits more from domestically generated and
available knowledge than from foreign knowledge. The patent variable - a proxy
for the generation of new codified knowledge - is only in rare cases
significant. Thus, tacit knowledge - researchers and students - plays a more
important role in regional growth than codified knowledge.
The results of this extended model show that the results discussed above
remain stable - not alone regarding their signs but also regarding the
coefficients - showing that they are independent of interregional knowledge
spillovers[16].
In light of the fact that most of the spillover variables are
insignificant in at least one sub-period it not possible to deduce a consistent
result as to the impact of knowledge spillovers. One minor insight arises as
both patent citation variables are positive and, at least in the later years,
also significant. Patent citations can also be viewed as an indirect indicator
of the presence of a significant research structure which generates the patents
and a pool of qualified inventors that are involved in the respective research.
Therefore, this positive impact can be viewed as a sign that a better research
system and better legislation regarding practical research will be beneficial
for the economic development of a region.
Finally, as the spatial term remains highly significant even though not
all of the spatial interactions are covered by knowledge spillovers. Especially
since the importance of the spillovers is rather marginal there are more
important links between the regions besides knowledge flows that have not been
explicitly included into the model.
Conclusions
The present study analyzed economic growth dynamics across the regions
of the Russian Federation. Besides proving the path-dependency of Russian
economic growth on a regional level, as well as its dependence on oil and gas,
it has been shown that knowledge does and always has played an important role
in the regional economic growth process. Here it is mostly researchers and to
some very minor degree the amount of students - as proxies of the stock of tacit
knowledge - that enhance economic growth while the amount of new patents - as a
proxy of the codified knowledge generated in each period - and most knowledge
inflows or outflows do not influence the economic development in any way.
In particular it has been shown that the impact of a doubling of the
output of the oil and gas sector does generate less additional economic growth
than a doubling of the number of researchers. Considering the future
development of the Russian Federation it is an important insight especially
since the result remains stable even during the transition years. It shows not
only that science intensive sectors are benefactor of economic growth but also
that the main dynamics of economic growth work comparably to Western European
economies. Russia can therefore learn from their growth strategies and
structural change programs to switch from being a resource-based to being a
knowledge-based economy.
[1]See Machlup
(1960) or Machlup (1962).
[2]While the
results in Netter and Megginson (2001) strengthen this argument, for the
Russian Federation Berkowitz and DeJong (2003) show that ownership has no
impact on firm performance; they instead highlight more the firms' distance
from Moscow, which in this study is implicitly included in the fixed effects.}
[3]It would be more
suitable to include a variable like the Corruption
Perception Index, or the ICRG index
of corruption advocated by Kim (2010) or the Bribe Payers Index advocated by Ofer (2010); however, they are not
available on a regional level for a continuous span of years. Note as well the
arguments by Brown and Shackman (2007) who link
corruption and the long-term development of the GDP per capita and a continuing
deterioration of law and order.
[4]Refer for the
resource curse hypothesis to Auty (1993), for example.
[5]For the Russian
Federation Popov (2001) for example stresses the importance of the level of
export shares for the regional performance.
[6]See Doehrn and
von Westernhagen (2003) as one article that stresses the importance of FDI for
growth in transition economies.
[7]See Blundell and
Bond (1998) and Bond et al. (2001) who consider the Blundell-Bond estimator to
be superior to the Arellano-Bond estimator in a growth related context.
[8]In total 80
regions have been considered. The three sub-regions of the Tyumen Oblast have
been jointly considered as well as Archangelsk and the Nenetsia Autonomous
Okrug.
[9]For reasons of a
better readability the coefficients for the openness variable has been multiplied
by the factor thousand.
[10]In other words raising the number of researchers by one percent leads to
rise in GDP by 0.32% while a one percent rise in oil and gas output only leads
to a rise in GDP by 0.05% to 0.07% - considering that researchers are measured
absolutely and production of oil and gas in thousand tons.
[11]As only a
version of the Patstat database from early 2008 has been available,
comprehensive patent data has only been available up to 2006.
[12]An active patent
citation takes place if an inventor from the region under consideration cites
another patent, whereas a passive patent citation takes place if a patent is
cited of which one inventor is registered living in the region.
[13]An inventor
inflow is registered if a patent is granted with an inventor being listed as
living in the region who has been listed in a previous patent as living in a
different region. Inventor outflows are defined vice versa.
[14]All patent
citation as well as inventor flow variables are
calculated based on EPO patent data, thereby accounting only for internationally
important knowledge.
[15]For a better
readability the coefficients of the patent and the four spillover variables
have been multiplied by a factor of thousand.
[16]Only the students variable - before significant and positive in the
later years - becomes insignificant in the extended model.
Bibliography
Ahrend, R., “Speed of Reform, Initial Conditions, Political Orientation,
or What? Explaining Russian Regions’
Economic Performance”, DELTA Working Paper, 2002-10, 2002.
Auty, R.,
“Sustaining Development in Mineral Economies: The Resource Course Thesis”,
Routledge, 1993.
Berkowitz, D. and DeJong, D., “Policy Reform and Growth in Post-Soviet
Russia”, European Economic Review, 47, 337-352, 2003.
Blundell, R. and
Bond, S. “Initial Conditions and Moment Restrictions in Dynamic Panel Data
Models”, Journal of Econometrics, 87, 115-143, 1998.
Bond, S.; Hoeffler, A. and Temple, J., “GMM Estimation of Empirical
Growth Models”, CEPR Discussion Paper, 3048, 2001.
Brown, S. and
Shackman, J., “Corruption and Related Socioeconomic Factors: A Time Series
Study”, KYKLOS, 60, 319-347, 2007.
Doehrn, R. and von Westernhagen, N., „The Role of Foreign Direct
Investment in Transformation“, In Lane, T; Oding, N. and Welfens, P.J.J.
(Eds.), “Real and Financial Economic Dynamics in Russia and Eastern Europe”,
251-278, Springer, Heidelberg, 2003.
Guellec, D. and
van Pottelsberghe de la Potterie, B., “From R&D to Productivity Growth: Do
the Institutional Settings and the Source of Funds of R&D Matter?”, Oxford Bulletin of Economics and Statistics, 66, 353-378,
2004.
Kim, M.J., “Corruption and Economic Growth”, EACGS Conference, 2010.
Lichtenberg,
F.R. and van Pottelsberghe de la Potterie, B., „International R&D
Spillovers: A Comment“, European Economic Review, 42, 1483-1491, 1998.
Machlup, F., “The Supply of Inventors and Inventions”,
Weltwirtschaftliches Archiv, 85, 210-254, 1960.
Machlup, F., “The Production and Distribution of Knowledge in the U.S.”,
Princeton University Press, Princeton, 1962.
Netter; J. and
Megginson, W., “From State to Market: A Survey of Empirical Studies on
Privatization”, Journal of Economic Literature, 39, 321-389, 2001.
Ofer, G., “Twenty
Years Later and the Socialist Heritage is Still
Kicking”, UNU-WIDER Working Paper, 59, 2010.
European Patent
Office, “Patstat Database”, http://www.epo.org, 2007.
Popov, V., “Reform Strategies and Economic Performance of Russia’s
Regions”, World Development, 29, 865-886, 2001.
Rosstat, “Regiony
Rossii”, http://www.gks.ru, 2012.
Surinov, A., “Voprosy Kolichestvennoj Otsenki Mezhregionalnykh Indeksov
Tsen”, Ekonomicheskij Zhurnal VSHE, 4, 604-613, 1999.
World Bank, “World
Development Indicators Database”, http://www.worldbank.org, 2012.
Appendix
A.1 Abbreviation of Variables
The abbreviations used in the presentation of the econometrical results
are summarized in the following table. It also contains the data sources of the
variables.
Table 5: Abbreviation of
Variables
*Jens K. Perret - European Institute for International Economic Relations at the University of Wuppertal
Rainer-Gruenter-Straße 21 42119 Wuppertal, Germany. Tel.: +49 202 439 3174 e-mail: perret@wiwi.uni-wuppertal.de
© 2010, IJORS - INTERNATIONAL JOURNAL OF RUSSIAN STUDIES