CC BY
19
The application of models of panel data analysis for the estimation of the volume of innovative goods, works and services in the Russian Federation
M. M. Tsvil1, V. E. Shumilina2
1Russian customs academy (Rostov affiliate), Rostov-on-Don;
2 Don State Technical University, Rostov-on-Don.
Abstract: This article deals with the carrying out of econometric analysis by means of panel data models (pooled model, fixed effects model) of the volume of innovative goods, works and services in the Russian Federation (RF) from 2010 up to 2014 years. The dependence has been revealed between the volume of innovative goods, works, services and such explanatory factors as the number of personnel involved in scientific research and development; internal expenses for the scientific researched and development of the Russian Federation; used advanced manufacturing technologies; coefficient of inventive activity; innovation activity of organizations.
Keywords: econometric analysis, panel data, innovation, science, integrated model of panel data (pooled model), a model of panel data with fixed effects (fixed effects model).
1. Introduction
In recent years, in terms of sanctions policy from Western countries more and more attention in the Russian Federation (RF) is paid to the creation and sale of innovative goods and to the provision of innovative works and services.
As is known, innovations are the central factor of production and productivity growth. The emergence of innovations is directly connected with the development of science. Science provides the economy and society with highly qualified personnel, revolutionary technological solutions and new technical ideas for everyday practical usage.
However, according to the report of the Russian Association of the science promotion (RASP) about the state of science in the Russian Federation, presented in 2013, the current state of the Russian science according to a number of objective indicators is characterized as catastrophic. One of the main reasons named was the organizational reason, i.e. the lack of a distinct state strategy in scientific and technical sphere. The problems of a lower level are arisen from this main problem, e.g. the rapid delay of Russia in the global field of scientific researches, the
problem of reproduction of qualified scientific and engineering personnel, the problem of introducing of scientific and technical innovations into the production and their incorporation into the chain of sectoral and intersectoral economic connections (bringing to the final product and to the final consumer); hence the problem of low share of small and medium enterprises in the structure of high-tech production and low share of industries with high added value in the GDP structure. Consequently, there is no large-scale demand for scientific developments by private companies and that factor exacerbates all the above-described problems and leads to the development of negative tendencies in self-reinforcing spirals.
At the St. Petersburg international economic forum on June 16-18, 2016 at the session «Big challenges - the stimulus of science development» the participants of which were the leading scientists and organizers of domestic and foreign science, the Chairman of the Board of the managing company «RUSNANO» A. Chubais mentioned that «Science itself has not been demanded by the production, science disturbs production as it requires big changes from any production». In his opinion, innovative economy should become the connecting link between science and production: «We need not expect that the production will create a request for science. It is needed to consider the chain of three elements: science - innovation economy - production. The demand for science from the side of manufacturers can be provided by innovative economy».
At that session, the member of the Expert Council under the Government of Russian Federation, Ekaterina Shapochka, pointed out the large data amount in the modern world, which is difficult to process by means of traditional ways; the Chairman of one of the world's largest scientific publishers, Elsevier B. V., Ionsuk Chi, noted that the value of data was increasing when they were structured, and even more - when analyzed.
The actuality of this topic determined the purpose of the investigation described in the given article.
2. The aim and methods of the investigation
The object of study is the volume of innovative goods, works and services in the Russian Federation (RF) from 2010 to 2014 (The data of Federal Statistics Service[1]). The objective of the investigation is to build adequate and substantial models to examine the true cause-effect relations between the volume of innovative goods, works, services in Russia (Y) in million rubles, and the explanatory variables: X - number of employees involved in scientific research and development (person); x2 - internal costs on scientific investigation and development of the Russian Federation in million rubles; X3 — used advanced manufacturing technologies (units); x - coefficient of inventive activity (number of domestic patent applications for inventions submitted in Russia per 10 thousand persons of population); x - innovative activity of organizations (specific gravity of organizations in % carrying out technological, organizational and marketing innovations in the reporting year, in the total number of the surveyed organizations) in eight Federal districts.
The experience of the econometric methods is of great interest in terms of uncertainty. Econometrics has become a powerful tool of economic research, especially in recent years as a result of the development of computer systems and special applied programs. The constantly complicated social and economic processes have led to the necessity of econometric analysis usage. The study of these processes is carried out by means of econometric models. The majority of new research methods are based on the econometric models, concepts and techniques. The application of econometric methods towards the real Russian statistical data will allow to understand deeper the goals and objectives of the state economic policy (or the company) and also to learn how to evaluate the results of this policy [2 - 6]. One of the possible solutions is the usage of the panel data. The application of estimation methods of the paneldata to solve the formulated task
seems to be reasonable as the temporary numbers data for Russia on the whole are insufficient to obtain good parameter estimations.
The panel data are those data, which contain statistical information about one and the same object set about for the number of consecutive time periods. Due to the structure the panel data allow to construct more flexible and substantial models and obtain answers to the questions that are unavailable only in the framework of models based on spatial data, for example. For economists, the panel data are valuable because it appears a possibility to consider and analyze the individual differences between the economic units that can't be done within the framework of standard regression models. Panel data models allow us to obtain more accurate estimated parameters [7 - 10].
3. The pooled models construction
In our case, the econometric analysis is based on panel data for the eight regions in the period from 2010 to 2014. The source of these data was the statistical data of the State Federal Service.
Table № 1.
Panel data in the Federal districts of the Russian Federation in the years 2010-2014
Federal Time y X 2 X з X4 X 5
districts (year)
1. 2010 290 757.6 381795 288960.0 68945 3.8 8.6
Central
2011 480 327.4 380363 331758.9 63078 3.27 10.2
2012 938 153.2 373461 369069.5 62796 3.71 10.9
2013 1 164 102.4 375087 398597.2 60829 3.77 10.7
2014 1 091 170.3 381047 447161.2 65591 3.0 10.9
2. North- 2010 120 105.5 95826 70737.3 17920 1.66 9.4
west
2011 196 049.1 97221 81504.9 19308 1.87 11.2
2012 298 020.1 97710 100002.7 18840 1.67 11.0
2013 409 750.4 95674 108026.7 19697 1.55 10.7
2014 354 113.0 96726 118612.3 20840 1.57 10.3
3. South 2010 86 558.4 28109 13027.3 7623 1.2 7.5
2011 59 811.8 27738 15906.0 7394 1.07 6.5
2012 51 801.6 23964 18618.0 7848 1.14 7.4
2013 70 281.9 24263 19987.0 8290 1.16 7.2
2014 102 845.3 25361 29274.3 9580 1.12 7.7
4. North- 2010 27 682.6 6053 2639.8 3194 2.01 6.2
Caucasian
2011 31 941.8 8585 4017.7 1993 1.24 5.2
2012 27 010.1 7188 3448.1 1833 1.61 6.4
2013 23 889.8 6330 3695.5 2113 1.74 5.9
2014 27 961.5 6628 4197.3 2215 0.71 6.5
4. North- 2010 27 682.6 6053 2639.8 3194 2.01 6.2
Caucasian
2011 31 941.8 8585 4017.7 1993 1.24 5.2
2012 27 010.1 7188 3448.1 1833 1.61 6.4
2013 23 889.8 6330 3695.5 2113 1.74 5.9
2014 27 961.5 6628 4197.3 2215 0.71 6.5
5. 2010 545 954.9 116285 74942.4 57394 1.38 12.3
Privolzhs
kiy
2011 781 944.9 111579 91012.1 55822 1.5 12.7
2012 950 604.8 114204 109155.0 54976 1.55 11.9
2013 1 128 642.7 114013 114194.6 57076 1.49 11.7
2014 1 179 545.3 107656 126552.5 59643 1.36 11.4
6. Ural 2010 109 584.6 42672 29441.8 35596 0.96 11.5
2011 179 708.9 43586 34408.9 30323 1.05 11.5
2012 148 696.2 43879 40420.2 31962 1.03 10.6
2013 189 234.1 44382 45167.0 31217 1.03 9.6
2014 169 373.1 45037 48800.0 29617 0.91 8.9
7. 2010 46 890.0 53024 33870.0 16335 1.25 8.2
Siberian
2011 88 866.0 52794 40713.4 15079 1.25 8.8
2012 117 118.0 52685 47011.7 15897 1.29 8.5
2013 151 362.7 53769 47666.3 16643 1.23 9.1
2014 186 025.2 54151 58435.9 18063 1.13 8.8
8. Far- 2010 16 178.9 12776 9758.7 5589 0.89 8.6
Eastern
2011 288 090.7 13407 11104.7 6595 0.89 11.2
2012 341 501.1 13227 12144.6 5810 0.93 10.8
2013 370 602.1 13227 12144.6 6801 1.04 9.5
2014 468 731.8 13204 13714.3 6956 0.95 8.9
For the econometric models with panel data the empirical analysis begins with the choice between models with a common effect (pooled model) and models with fixed effects (fixed effect model).
Pooled model - is the usual linear regression model, which in matrix form looks like this:
y = XP + S, (1)
for the coefficients estimation of which the ordinary least-squares method (LSM) may be used.
In our case the built model on the basis of the panel data the pooled model (1) has the following form:
y = -451151421 - 7.794X1 + 6.52X2 + 12.625X3 + 176997.2X4 + 361112243X5 (2)
The absence in panel structure data and the possibility to receive consistent and effective assessments by means of the pooled sample with the help of LSM is formulated as a null hypothesis in F-test.
The determination coefficient of the given model (2) is R2 = 0.82. The dependence of y from x^, x2, X3 , x^, x^ is characterized as close in which 82% of the volume variations of innovative goods, works and services is determined by the variation considered in the model factors: the number of personel involved in scientific research and development; internal expenses for the scientific research and development throughout the Russian Federation; the used advanced manufacturing technologies; inventive activity coefficient; innovative activity of organizations in eight Federal districts.
The regression equation according to F-Fisher criterion is statistically significant. Let's give the main parameters of the model (2):
Table № 2.
Main parameters of the model (2)
Index Coefficient t-statistics P-meaning
Ао -451151.421 -1.961548322 0.058045124
А -7.794 -4.629470975 5.16537E-05
А 6.520 5.352415239 5.98749E-06
Аз 12.625 4.383314752 0.000106584
А4 176997.200 1.621723842 0.114100742
А5 36112.243 1.735211339 0.091760779
As is seen from the table 2 the coefficients of the model (2) are mainly all statistically significant according to Student's test with a significance level a = 0.1 and the number of degrees of freedom equal to 34 ( ta = 1.68 ).
It is evident from the given model that the growth of volume of innovative goods, works and services is certainly positively influenced by a growth of the coefficient of inventive activity (X4) and innovative activity of organizations (X5). With the values X4 and X5 equal to 1 and X, x2 , X3 equal to 0, the volume of innovative goods, works and services is equal to 86958.022 million rubles.
With a significance level a = 0.05 the coefficients p0, p and p5 are not significant according to Student's criterion. It is explained by the fact that only the Central Federal district has simultaneously significant coefficients of inventive activity and innovative activity of organizations.
Let's compare the pooled model with the fixed effect model allowing to estimate the influence of values of quantitative attributes in each region separately.
In matrix form the model looks like this:
y = xp + za + s, (3)
where Z = {z1 , Z2,..., Z8).
The model (3) presupposes the introduction of dummy variables Z = Z, Z,.., Z) for the sample objects. The coefficients in dummy variables will give the valuations of the individual effects. Fixed effects model ensures the guaranteed receiving of unbiasedand consistent valuations. In our case, Z takes the value 1 for the data of the Central Federal district, for other districts it is 0; z takes the value 1 for the data of the North-Western Federal district and for the other districtsit is 0; z takes the value 1 for the data of the Southern Federal district and for other districts it is 0 etc.
Built on the basis of the panel data the fixed effects model has the following
form:
y = 5534088.579 - 19.437X + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5 + 0 • Z -- 4159517.582Z --5207747.819Z3 - 5562511.087Z4 - 3182346.686Z5 - 4861327.279Z6 -- 4797898.101Z7 - 5173066.687Z8 (4)
The coefficient of determination of the model (4) R2 = 0.96 is statistically significant. From the constructed model it is followed that the factor y is mainly positively influenced by the internal costs ( x ), innovative activity of organizations ( X4 ), the coefficient of inventive activity ( X5 ).
So, for the Central Federal district the model (4) has the following form: y = 5534088.579 - 19.437X + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5, (5)
where ^ = 1, z2 = z3 = z4 = z5 = z6 = z7 = z8 = 0. For The North-West district the model has the following form: y = 1374579.997 - 19.437X + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5 (6) where z2 = 1, z = z3 = z4 = z5 = z6 = z7 = z8 = 0. For the South district the model has the following form: y = 326340.76 -19.437X1 + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5 (7)
where z3 = 1, z = z2 = z4 = zs = z6 = z7 = z8 = 0.
For The North Caucasus district the model has the following form: y = -28422.508 - 19.437X + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5, (8)
where z4 = 1, z = z2 = z3 = z5 = z6 = z7 = z8 = 0.
For the Privoljski district the model has the following form: y = 2351741.893 - 19.437X + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5, (9)
where z5 = 1, z = z2 = z3 = z4 = z6 = z7 = z8 = 0.
For the Ural district the model has the following form: y = 672761.3 -19.437X1 + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5 , (10) where z6 = 1, z = z2 = z3 = z4 = z5 = z7 = z8 = 0.
For The Siberian district the model has the following form: y = 736190.478 -19.437X1 + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5 , (11) where z7 = 1, z = z2 = z3 = z4 = z5 = z6 = z8 = 0.
And for the Far -Eastern district the model has the following form:
y = 361021.892 -19.437X1 + 6.26X2 - 1.34X3 + 106182.495X4 + 2749.936X5 (12) where Z8 = 1, Z = Z = Z = Z = Z = Z = Z = 0.
4. Conclusion
Thus, the practical significance of the models (5-12) is that it will allow to predict and calculate the volume of innovative goods, works and services taking into account the number of personnel involved in scientific research and development; internal costs for the research and development; used advanced manufacturing technologies; coefficient of inventive activity and innovation activity of organizations in each of the above mentioned regions of the Russian Federation.
Notes: In the course of writing of this article on the site of Federal State Statistics (Official Rosstat Statistics [1] ) there have appeared the actual data for the year 2015, according to which, the meaning Y, for example, for the Central Federal district was 1491536.1 million rubles, and the prognostic value according to the model (5) in accordance with the data of 2015 was 1562198.7678 million rubles, that evidences about the high prognostic quality.
On the basis of the above mentioned it is possible to make a conclusion that the innovation development has the great influence on the economic development of the Russian Federation under present-day conditions. The volume increase of the innovative goods, works and services is positively influenced by the innovation activity of organizations, inventive activity and domestic costs on research and development.
References
1. Dannye Federal'noy sluzhby gosudarstvennoy statistiki. URL: gks.ru/wps/wcm/connect/rosstat_main/rosstat/ru/statistics/science_and_innovations /science/.
2. Tsvil' M.M., Kolesnikova I.V. Inzenernyj vestnik Dona (Rus), 2016, №4 URL: ivdon.ru/ru/magazine/archive/n2y2016/3591.
3. Tsvil' M.M., P'yanova Yu.S. Aktual'nye problemy gumanitarnykh i estestvennykh nauk, 2016. № 6-2, pp. 96-99.
4. Tsvil' M.M., Shumilina V.E. Inzenernyj vestnik Dona (Rus), 2014, №4 URL: ivdon.ru/ru/magazine/archive/ N4y2014/2555.
5. Tsvil' M.M., Shumilina V.E. Inzenernyj vestnik Dona (Rus), 2015, №2, URL: ivdon.ru/uploads/article/pdf/ivd_10_Tsvil1.pdf_ebd0610868.pdf.
6. Tsvil' M.M., Shumilina V.E. Inzenernyj vestnik Dona (Rus), 2014, №1 URL: ivdon.ru/ru/magazine/archive/n1y2014/2241.
rrl
7. Baltagi B.H. Econometric Analysis of Panel Data. B.H. Baltagi. 3 Edition. Chichester: John Wiley &Sons, Ltd, 2005, 356 p.
8. Matyas L. The Econometrics of Panel Data. Fundamentals and Recent
rrl
Developments in Theory and Practice. L. Matyas, P. Sevestre (eds.). 3 Edition, 2008, Berlin, Springer, 955 p.
9. Ekonometrika: uchebnik dlya magistrov [Econometrics: the manual for masters], I.I. Eliseeva [i dr.]; pod red. I.I. Eliseevoy. M.: Izdatel'stvo Yurayt, 2012. 453 p.
10. Greene W.N. Econometric Analysis. W.H. Greene. 4th Edition. New Jersey: Prentice Hall, 2002, 272 p.
