Application of Econometric Methods and Neural Network Analysis in

Regional Sustainable Development Management Programs

Luisiena E. Puinko

and Elena V. Tolkacheva

Far-East Institute of management, branch of the Russian Presidential Academy of National Economy and Public

Administration (hereinafter RANEPA), Khabarovsk, Russia

Keywords: Neural Networks, Socio-Economic Processes, Public Administration.

Abstract: The digitalization of the economy in Russia is subject to management influence from Federal, regional, and

municipal Executive authorities. At the same time, there is a continuous search for methods and tools to

improve its effectiveness. This process is complicated, among other things, by the fact that trends at all levels

of economic digitalization management and mechanisms for its implementation in Russia currently remain

insufficiently studied. The use of classical statistical and econometric methods for assessing and predicting

socio-economic processes, both in Russia as a whole, and in individual regions and municipalities on its

territory, has proven itself well; and at the same time, they are very time-consuming, and the result obtained

through the use of statistical or econometric methods of analysis is obtained with a certain time delay; and at

the time of its receipt, it does not correspond to the stated goals of the study. Then econometric analysis and

statistical methods should be replaced by a tool that will allow you to get results faster with no less quality,

and use it in a timely manner when implementing and correcting management tasks. One of the directions of

development of new tools for analyzing many processes of dynamics, stochastic processes, and systems with

big data is artificial intelligence, or information systems based on it. In the conditions of incomplete

information about the socio-economic processes of any region, statistical methods of assessment can even

give an erroneous result, which can provoke fatal management errors. To minimize forecasting estimates and

optimize analysis and evaluation procedures, it is necessary to rely on modern, new and effective methods for

analyzing stochastic processes.

1 INTRODUCTION

The regional system is a complex mechanism for the

functioning of objects and their relationships. The

development of any socio-economic system (regional

and industry level) is a process of interaction of trends

and patterns arising from the characteristics of this

system, which is internally contradictory and

dynamic (Decree Of the government of the Russian

Federation No. 313, 2014). The study of regional

socio-economic systems is complicated by at least

two fundamental difficulties. First, socio-economic

systems are complex systems that depend on a very

large number of variables (Decree of the President of

the Russian Federation No. 204, 2018). And,

secondly, the behavior of such systems is difficult to

https://orcid.org/0000-0002-8938-130X

https://orcid.org/0000-0003-1304-809X

formalize and predict (Federal state statistics service,

2019).

1.1 The Basic Methods in the System of

Regional Management

In General, it should be noted that regional analysis

methods are understood as a set of tools for analyzing

the location and development of regional systems.

Among the main methods for regional studies are

highlighted in the comparative-geographical method,

the method of synthesis and analysis and iterative

method, method of economic and mathematical

modeling, the balance method, program and target

method, mapping method, index method, methods of

sociological research method, the zoning method of

development potential, the method priorities

116

Puinko, L. and Tolkacheva, E.

Application of Econometric Methods and Neural Network Analysis in Regional Sustainable Development Management Programs.

DOI: 10.5220/0010587001160122

In Proceedings of the International Scientiﬁc and Practical Conference on Sustainable Development of Regional Infrastructure (ISSDRI 2021), pages 116-122

ISBN: 978-989-758-519-7

(Zubarev, 2017; Zhadan, 2018; Sudarushkina and

Stefanova, 2017; Chernova, 2019; Shvab, 2016;

Digital economy-national projects-Khabarovsk

territory, 2020; Official portal "Ministry of digital

development, communications and mass

communications of the Russian Federation", 2020).

1.2 Assessment of Individual Indicators

of Socio-economic Development in

the Digital Economy Contour

In terms of the use of information and communication

technologies, in General, the regions of the far

Eastern Federal district reach the national average,

while the remoteness of the far Eastern Federal

district from the Western regions of Russia plays an

important role. on the one hand, the use of the Internet

and home computers in households reaches an

average of 70%, while online purchases make up

28.6%, which is associated with the delivery of these

goods to remote areas (i.e., a poorly developed

logistics network for delivering online purchases to

the population) (Federal state statistics service, 2019).

The use of public services on the Digital government

portal reaches 47% in the far Eastern Federal district,

which, according to the authors, is due to insufficient

coverage of public services that can be obtained by

the population in electronic form, and MFC exist and

do not work in all hard-to-reach areas of the far

Eastern Federal district (or their work is not

profitable) (Decree Of the government of the Russian

Federation No. 313, 2014; Decree of the President of

the Russian Federation No. 204, 2018).

2 PROBLEM STATEMENT

The authors of the study analyze individual socio-

economic processes related to the development of the

digital economy in the region. We tried to evaluate

individual indicators of socio-economic development

using classical methods of econometric analysis, and

also selected various models to implement data

analysis as accurately as possible. For the purpose of

the study, the further use of the obtained results of the

found model for forecasting and adjusting

management programs was laid. However, there was

a complication that the time spent by researchers to

search for and evaluate the results of the selected

model was large. The time interval for obtaining a

ready-made assessment and the possibility of using it

to adjust socio-economic development measures

overlapped with the time for calculating and selecting

the desired model. At the time of receiving the results

of the assessment of socio-economic development

processes in the region in the digital economy

contour, these results were already outdated. The

authors tried to determine whether classical methods

of econometric analysis are suitable for assessing

complex systems of socio-economic development in

the region in modern conditions. At the same time, I

would like to emphasize that econometric methods

and models have proven themselves well for

analyzing and evaluating relatively simple

dependencies: when considering the relationship of

several factors that affect the result, there are not

many endogenous and exogenous variables in the

model; each individual factor included in the model

does not represent a complex system of interaction of

additional variables. However, socio-economic

processes in the region are always a multi-factor

complex system, in which individual variables, both

endogenous and exogenous, can be revealed, again,

as a complex system. For example, the standard of

living of the population depends on nominal accrued

wages, expenditure patterns of households,

unemployment rate, availability of medical care,

structure, household income, housing affordability,

crime rates in the region, etc., each listed variable of

influence is a complex system dependent on many

other factors, and are interrelated. Formal work is like

unemployment rate and crime rates (crime rate and

unemployment rate are directly proportional), the

same processes affect training of the population, the

possibility of retraining, the level of culture, family

values etc.

Any econometric analysis involves the processing

of statistical information previously collected either

through official (state) services, such as regional

divisions of Rosstat, or through specialized

representative surveys, which involve the

involvement of both state and non-state funds and

institutions of socio-economic analysis (Federal state

statistics service, 2019).

At the same time, the econometric models

themselves are divided into spatial models and

temporal models (trend analysis models). Spatial

models assume an assessment of socio-economic

processes at a given time. Time series-construction of

the development trend of the socio-economic process

under study for a number of consecutive moments in

time. Both in the first and in the second case the

forecast values can be calculated quite successfully

based on the found model; in both cases, the sample

of data used to build an econometric model should not

only be representative of the quality of the collected

statistical data, but also in terms of volume: less than

Application of Econometric Methods and Neural Network Analysis in Regional Sustainable Development Management Programs

117

thirty observations are usually not used for qualitative

econometric analysis. For a qualitative analysis of

time series, the sample size must be a multiple of.

Many socio-economic processes are stochastic,

which complicates the choice of tools for analysis

(Federal state statistics service, 2019; Piunko, 2019).

2.1 Regional Management of

Socio-economic Processes as a

Form of Spatial Analysis Models

Regional management of certain socio-economic

processes is a rather complex system, which from the

position of econometric analysis can be theoretically

justified either by a system of econometric equations

(according to the authors), or by a time trend. In our

opinion, it is not correct to apply the multiple

regression model here, since regional management is

not a single factor under study, but a set (complex) of

management factors in various sectors of the

economy and the social sphere. Thus, you can design

"regional management" as a model of the form ŷ =

f(x1; x2; x3; x4;...; xn ), where xn are individual

socio-economic indicators, does not seem to be

correct for research (the multiple regression model is

not applicable here, it can only be applied to

individual processes in management).

In our opinion, multiple regression and correlation

can only be applied to certain socio-economic

processes, such as: the convenience of registering a

child in a preschool educational institution, including

through electronic public services, the convenience of

providing tax reporting, including through network

communication channels, individual indicators of

electronic interdepartmental interaction, the

availability of medical services, including using

electronic public services, etc.

2.2 Analysis of Socio-Economic

Processes in the Region through

Systems of Econometric Equations

Systems of econometric equations assume a

preliminary knowledge-intensive econometric

analysis, the purpose of which is to select independent

factors in the system, which are then built up in the

system and distributed in its equations.

At the same time, the selected factors themselves

cannot correlate with each other, and there should be

no duplicate factors in the system. After constructing

a model system of econometric equations describing

state regulation, it is necessary to determine the

condition of identifiability for it, and then calculate

the parameters of this system.

Such a study is very time-consuming, and given

the current dynamics of regional management

processes, we assume that a theoretically constructed

model of a system of econometric equations

describing regional management as a system of

interrelated factors (the most significant at the

beginning of the study), at the time of evaluating its

quality and calculating forecast values (i.e., at the

final phase of the study) – will give an outdated result,

which will be somewhat distorted forecast values. In

other words, the laws of statistical analysis, which

compare the summary characteristics of aggregates of

objects and phenomena related to the life of society

in different periods of time, also work here.

2.3 Methods of Time Series Analysis

for Forecasting Socio-economic

Processes in the Region

Time series analysis, primarily the study of economic

processes through VAR models, which is a special

case of a system consisting of simultaneous

equations, analysis through ARMA and ARIMA

models, analysis of socio-economic processes as

difference-stationary series (DS-series, Difference

Stationary) or autoregressive fractional integrated

moving average model (autoregressive fractionally

integrated moving average – ARFIMA, which

involve collecting a very large number of

observations from 800 or more), the use of

autoregressive models with distributed lags (ADL

models) - as a kind of dynamic models, also involves

collecting a large range of data under study. Bayesian

time series models, which involve hierarchical

modeling and are used when information is available

at several different levels of observable quantities, are

convenient for hierarchical analysis, and in some

cases for multidimensional analysis; but in the

context of research on public management of socio-

economic processes, including the digitalization of

the economy, it is not convenient and takes a lot of

time to build them (first of all, a large range of initial

statistics - data collection for analysis).

Analysis of sales data indicators for the digital

economy; was chosen a model with fixed effects

analysis of panel data, model selection was due to the

goal setting criteria of the target values of the national

project "Digital economy", and, as well, the statistical

data for analysis.

ISSDRI 2021 - International Scientiﬁc and Practical Conference on Sustainable Development of Regional Infrastructure

118

3 RESEARCH QUESTIONS

The study was supported by the far Eastern Institute

of management, a branch of the Russian state

Academy of national economy and public

administration under the President of the Russian

Federation and the government of the Khabarovsk

territory. The research was carried out using neural

network analysis of individual indicators of digital

economy development in the region, assessment of

measures related to the management of socio-

economic processes. As part of the study, a survey of

experts was conducted in the number of 45 people

who are representatives of business entities, regional

authorities, the scientific community, and public

organizations. During the survey, the task was set: to

determine the conceptual validity of the created

neural network models, taking into account the

expected impact of digitalization of the economy on

the organizational and administrative activities of

government bodies.

3.1 The Importance of using Neural

Network Analysis in Management

as the Main Hypothesis of the

Study

In formulating the questions the expert survey was

based on the hypothesis that representatives of

business entities, regional authorities, scientific and

educational community, and public organizations

interested in the framework of their professional or

educational activities, as well as the implementation

of functions of public control in obtaining reliable

analytical information about the organizational and

managerial activities of the regional authorities on

digital economy development in Khabarovsk Region.

3.2 Evaluation of an Expert Survey on

the Use of Neural Network Analysis

in Management

When deciding about the implementation of neural

network analysis for analyzing management and

administrative activities of regional authorities in

Khabarovsk region on the development of the digital

economy must consider the lack of digital

competence of all potential categories of users:

researchers, faculty, staff, business representatives,

representatives of public organizations (Federal state

statistics service, 2019).

4 PURPOSE OF THE STUDY

The aim of the work was to provide theoretical

justification and develop a method for neural network

analysis of methods of regional management of

digitalization of the Khabarovsk territory economy,

as well as recommendations for its practical

application.

Currently, a number of management measures

have been taken in the Khabarovsk territory to

achieve digital integration of communications in

public administration, in the provision of state and

municipal services, as well as to create a modern

information environment in the production of goods

and services. At the same time, a number of issues

arise that directly affect the assessment of possible

socio-economic risks of making certain management

decisions at the regional level. The results of this

study will expand the theoretical understanding of the

methods of regional management of digitalization of

the economy in the Khabarovsk territory, as well as

test the methodology for studying the results of

regional management based on artificial intelligence.

5 RESEARCH METHODS

For the purposes of this study, the most interesting

were the normative and index methods. The

normative method is a method of substantiating

indicators of socio-economic development of the

region using pre-developed and legally established

norms and standards. Norms and regulations

represent the necessary base, scientific development

of regional economic forecasts, plans, programs,

technical and economic projects. As for the index

method, it is most often used to quantify the level of

specialization of socio-economic zoning. In this case,

the base year indicator is taken as a basis and the

growth or growth rates of indicators are calculated,

which should collectively show statistically

significant phenomena. These methods were used in

the analysis of individual indicators of the

development of the digital economy of the

Khabarovsk territory.

6 FINDINGS

In the current dynamics of socio-economic

development, there is a need to use other approaches

to collecting and processing statistical data based on

modern intelligent information technologies. They

Application of Econometric Methods and Neural Network Analysis in Regional Sustainable Development Management Programs

119

allow, firstly, to collect and process data faster,

secondly, their use is cheaper than traditional

methods, and thirdly, the result of these studies in

terms of validity and reliability will be higher than in

traditional studies.

6.1 The Inappropriateness of the Use of

Econometric Analysis in

Forecasting Complex Systems of

Socio-Economic Governance

Regional management in the field of digitalization of

the economy also involves consideration of the

management function as a system of management

factors in the digitalization of services to the

population, digitalization of interdepartmental

interaction, development of state information

systems, replacement of imported software with

similar domestic production (especially in the public

sector), digitalization and development of regional

market platforms, digital monitoring of individual

socio-economic processes (including improving

monitoring of public procurement through

appropriate on-line platforms), calculating

transaction costs, etc.

If the task for the researcher is just to evaluate a

separate control factor, or its separate process, then

multiple regression models or systems of econometric

equations can be considered as a fairly reliable and

convenient research tool.

Due to attempts to apply econometric analysis in

practice (primarily time series analysis) of individual

socio-economic processes; difficulties in finding

reference statistical data to include in the model under

study; difficulties with calculating the model

parameters, and most importantly, evaluating the

quality of the found model, which in terms of

indicators was not always suitable for further

forecasts and did not correctly assess the effects

obtained from certain management processes (public

administration activities to achieve the stated goals),

it was concluded that it is necessary to use other tools

in terms of evaluating and analyzing management

processes. Moreover, we considered tools and models

that could independently correct system errors within

themselves. Econometric and statistical analysis had

to be abandoned due to the above disadvantages.

Analysis of sales data indicators for the digital

economy; was chosen a model with fixed effects

analysis of panel data, model selection was due to the

goal setting criteria of the target values of the national

project "Digital economy", and, as well, the statistical

data for analysis (Official portal "Ministry of digital

development, communications and mass

communications of the Russian Federation", 2020).

Also, based on the results of sociological surveys,

they often build an econometric model of the studied

factor or group of socio-economic factors, getting

some error in the analytical model. Given that the

socio-economic processes taking place in the regions

of Russia are usually complex and multi-factorial, the

construction of econometric models for them is often

impractical.

6.2 Possibilities of using Information

Systems for Statistical Data

Analysis in Management Models

It should also be noted that modern domestic

developments of information technologies and

information systems provide partial automation of

operations in the state management of socio-

economic processes. They allow processing statistical

data on various socio-economic indicators, are

progressive, and flexible in integration. Also,

information technologies significantly reduce the

time of specialists in performing statistical and

economic analysis tasks. The development of

domestic software, modern information systems and

technologies are particularly in demand today, as they

are tools for integrating the entire state management

system at all levels into the digital economy.

6.3 Difficulties in using Neural

Network Technologies in the

Analysis of Public Administration

in the Khabarovsk Territory

When deciding about the implementation of neural

network analysis for analyzing management and

administrative activities of regional authorities in

Khabarovsk region on the development of the digital

economy must be considered insufficient digital

competences of all potential categories of users:

researchers, faculty, staff, business representatives,

representatives of public organizations. At the same

time, you can also take into account the importance

of using domestic software, which leads to two ways:

using a package of applications for statistical data

analysis with a built-in machine learning add-on, or

ordering software from a developer.

At the same time, the software being developed

must be able to perform statistical data analysis in

addition to machine learning capabilities.

ISSDRI 2021 - International Scientiﬁc and Practical Conference on Sustainable Development of Regional Infrastructure

120

6.4 The Main Content of the Expert

Survey on the Use of Neural

Network Analysis in Management

Experts were asked to evaluate the statements that

characterize the features of using neural networks to

analyze the regional management of digitalization of

the Khabarovsk territory economy. In particular, the

assessment was based on the following indicators:

 The need to use neural networks in analyzing

the development of the digital economy in the

Khabarovsk territory;

 Sufficiency of application of statistical

methods in the analysis of the activities of

regional authorities of the Khabarovsk

territory;

 Difficulties in using neural networks in

regional authorities due to the lack of

understanding of the features of applying the

neural network analysis methodology by

employees;

 The importance of using domestic software in

conducting neural network analysis of the

activities of the Khabarovsk territory

authorities for the development of the digital

economy;

 The potential of neural networks for

operational adjustment of the activities of

regional authorities for the development of the

digital economy in the Khabarovsk territory.

In formulating the questions the expert survey was

based on the hypothesis that representatives of

business entities, regional authorities, scientific and

educational community, and public organizations

interested in the framework of their professional or

educational activities, as well as the implementation

of functions of public control in obtaining reliable

analytical information about the organizational and

managerial activities of the regional authorities on

digital economy development in Khabarovsk Region.

6.5 Results of an Expert Survey on the

Use of Neural Network Analysis in

Management

Based on the responses received, we identified five

groups of experts, and the following results of the

expert survey were obtained.

The largest group of experts supports the

introduction of neural network analysis to analyze the

activities of regional authorities for the development

of the digital economy using domestic software.

The second group of experts indicates the

importance of using neural network analysis of

government activities in conjunction with statistical

methods of analysis.

The third group of experts supports the

introduction of neural network analysis to analyze the

activities of regional authorities in the development

of the digital economy, but at the same time

emphasizes the non-necessity of using domestic

software.

There is also a group of experts who do not

support the use of neural network analysis of

government activities, focusing on the use of

statistical methods.

And you can select a group of neutral experts.

In General, it can be noted that 63.5% of experts

support the need to use neural networks to analyze the

activities of regional authorities, having a difference

of views only in the tools for its implementation.

So 56% of experts noted that to analyze the

development of the digital economy in the

Khabarovsk territory, it is necessary to use neural

networks that can build complex predictive models.

52% of experts note that it is also necessary to

continue using statistical methods for analyzing the

activities of regional authorities.

48% of experts noted that the use of neural

networks by employees of regional authorities of the

Khabarovsk territory is currently difficult due to

insufficient knowledge of the neural network analysis

methodology.

At the same time, 52% noted that neural networks

will help quickly adjust the activities of regional

authorities for the development of the digital

economy in the Khabarovsk territory.

Thus, when deciding about the implementation of

neural network analysis for analyzing management

and administrative activities of regional authorities in

Khabarovsk region on the development of the digital

economy must be considered insufficient digital

competences of all potential categories of users:

researchers, faculty, staff, business representatives,

representatives of public organizations.

7 CONCLUSION

Thus, in this study, a theoretical model of

administrative and administrative methods of

regional management of digitalization of the

economy in the Khabarovsk territory was developed.

Information models of neural networks used for

analyzing administrative and administrative activities

of regional authorities of the Khabarovsk territory

also received their justification. Based on the

calculation of validity errors, their verification is

Application of Econometric Methods and Neural Network Analysis in Regional Sustainable Development Management Programs

121

evaluated. The main hypothesis of the study is that the

use of neural network models will eliminate the lack

of knowledge about the possible socio-economic

consequences of the use of various administrative and

administrative methods of regional management of

the digitalization of the economy in the Khabarovsk

territory; in the course of the study, the hypothesis

was confirmed.

This applied research allowed us to clarify the

existing knowledge about the possibilities of using

neural network models to analyze the results of

regional management of economic digitalization,

including in the Khabarovsk territory.

ACKNOWLEDGEMENTS

Within the framework of the agreement on the Grant

of the Ministry of Education and science of the

Khabarovsk territory dated August 21, 2020 No.

65C/2020 “Neural network analysis of methods of

regional management of digitalization of the

Khabarovsk territory economy”.

REFERENCES

Decree Of the government of the Russian Federation No.

313 of 15.04.2014 "On approval of the state program"

Information society (2011-2020)"». (2014).

https://digital.gov.ru/ru/documents/4137/.

Decree of the President of the Russian Federation of

07.05.2018 No. 204 "On state goals and strategic

objectives of the development of the Russian

Federation for the period up to 2024" (2018). URL:

http://www.kremlin.ru/acts/bank/43027/page/1.

Federal state statistics service: information on the socio-

economic situation in Russia, January-August 2019.

URL: https://rosstat.gov.ru/storage/mediabank/oper-

08-2019.pdf .

Federal state statistics service: information on the socio-

economic situation in Russia, January-August 2019.

https://gks.ru/storage/mediabank/oper-08-2019.pdf .

Piunko, L.E. (2019).Directions of development of digital

economy and information society in Russia. 1st

International Scientific Conference on Modern

Management Trends and the Digital Economy - From

Regional Development to Global Economic Growth

(MTDE) -Yekaterinburg, RUSSIA, volume 81 pages

150-153. https://www.atlantis-

press.com/proceedings/mtde-19/125908808.

Zhang, Y.A., Yan, B.B.; Aasma, M. (2020). A novel deep

learning framework: Prediction and analysis of

financial time series using CEEMD and LSTM. Expert

systems with applications, volume 159.

https://www.sciencedirect.com/science/article/pii/S221

491471930501X?utm_source=TrendMD&utm_mediu

m=cpc&utm_campaign=Defence_Technology_Trend

MD_1 .

Zubarev, A. E. (2017). Digital economy as a form of

manifestation of regularities in the development of the

new economy. Vestnik TOGU, 4 (47): 177-184.

Zhadan, I. E. (2018). Social risks in the digital economy.

Humanitarian scientific journal, 20-26.

Sudarushkina, I. V., Stefanova, N. A. (2017). Digital

economy. ANI: Economics and management, 1(18):

182-184.

Chernova, V. Yu. (2019) Russian consumer in the digital

economy. Discussion, 2 (93): 38-41.

Shvab, K. (2016). The Fourth industrial revolution.

"Eksmo", Top Business Awards.

http://ncrao.rsvpu.ru/sites/default/files/library/k.

shvab_chetvertaya promyshlennaya_revolyuciya

2016.pdf .

Digital economy-national projects-Khabarovsk territory.

(2020). https://www.khabkrai.ru/khabarovsk-

krai/Proekty/175469. (date of request: 08.10.2020)

Official portal "Ministry of digital development,

communications and mass communications of the

Russian Federation" (2020).

https://digital.gov.ru/ru/pages/statistika-otrasli/ .

ISSDRI 2021 - International Scientiﬁc and Practical Conference on Sustainable Development of Regional Infrastructure

122