=Paper=
{{Paper
|id=Vol-2922/paper001
|storemode=property
|title=Using neural network technologies to reduce information asymmetry
|pdfUrl=https://ceur-ws.org/Vol-2922/paper001.pdf
|volume=Vol-2922
|authors=Alexandr Kindaev, Alexandr Moiseev,Ekaterina Kolobova
}}
==Using neural network technologies to reduce information asymmetry==
https://ceur-ws.org/Vol-2922/paper001.pdf
Using neural network technologies to reduce information
asymmetry*
Alexandr Kindaev[0000-0002-3855-1970], Alexandr Moiseev[0000-0001-9534-2465] and Ekaterina
Kolobova[0000-0002-9569-2526]
Penza State Technological University, 1а/11, Baidukova Passage/Gagarina Street, Penza,
440039, Russian Federation
kindaev@penzgtu.ru
Abstract. The article discusses an approach to the analysis of static information
based on the use of neural network technologies. Kohonen maps were
constructed on the basis of data on the yields of the main agricultural crops in
four adjacent areas of the Volga region over 30 years. As a result, based on the
analysis of the maps obtained, three clusters were identified: a cluster with high,
medium and low yields for the main crops. Also, on the basis of a comparative
analysis, regions were identified for which similar results were observed for
different crops. The results obtained have a significant impact on the process of
making managerial decisions regarding the use of the agricultural insurance
tool.
Keywords: Kohonen maps, correlations, clusters, decision making.
1 Introduction
Information asymmetry is characterized by uneven access to the same information by
the parties of economic relations. It turns out that one party, when making
management decisions, has more opportunities to calculate potential risks or to assess
the amount of possible income. Greater awareness can be caused, for example, by the
level of professional knowledge, accumulated experience, knowledge of the specifics
of a product or service, as well as their hidden properties, etc. The presence of
information asymmetry negatively affects both the economy of a particular industry,
and as a whole, and can also lead to the failure of the market in a particular area, as
there is a shift in the optimal price of a product or service [6; 11]. The party that owns
more information gets a significant share of the profit at the expense of less risk,
while the other party suffers losses due to the lack of similar access to information,
which may ultimately lead to bankruptcy or refusal to interact with a specific product
or service [1; 3].
*
Copyright c 2021 for this paper by its authors. Use permitted under Creative Commons License
Attribution 4.0 International (CC BY 4.0).
The problems of studying and overcoming the asymmetry of information are dealt
with by both domestic and foreign specialists [2; 5]. So, in the article [10] deals, price
signals and efficiency in markets with information asymmetry, where there are two
situations: uninformed sellers set high prices, and uninformed buyers are ready to buy
at low prices, and when uninformed sellers set low prices, and uninformed buyers are
willing to buy at high prices. In the article, they show that the largest number of
transactions is observed in the second situation. The article [9] examines the foreign
exchange market, where the authors find convincing evidence of heterogeneous
excellent information on agents, time and currency parameters, which is consistent
with the theory of asymmetric information and fragmentation of the OTC market. A
trading strategy based on constant price influence, capturing the risk of asymmetric
information, gives high returns even after taking into account the risk, transaction cost
and other general risk factors described in the foreign exchange market literature.
There is also research on the awareness of management decision makers. The article
[7] states that there is a perception asymmetry among politicians from urban to village
levels, which leads to the fact that disaster management programs do not become a
priority when using rural funds.
2 Materials and methods
To overcome uncertainty in individual areas, it is necessary to correctly identify
groups of similar objects, which will allow identifying their unique characteristics. In
the scientific literature, three methods of object classification are traditionally
distinguished: hierarchical, faceted, descriptor. However, in recent years, neural
network technologies have been increasingly used in the study and clustering of
objects. One way to reduce uncertainty and reduce information skewness is self-
organizing Kohonen maps, which is an unsupervised learning model designed for
applications where it is important to maintain a topology between input and output
spaces. Figure 1 shows the structure of the Kohonen map, on which the input and
output levels are present. It is also worth noting that all input neurons are connected to
all neurons in the output layer.
In this network, neurons learn on their own, without anyone's control, since
competition is at the heart of the network and learning. Only vectors are fed into the
input, which make up the training sample. The activation of a neuron is calculated as
the distance between its weight vector W and the input pattern X. Distance can be
defined as dot product or Euclidean distance. Two training rules may apply: Winner-
Take-All (WTA) where only the winner's weights change, or Winner-Take-Most
(WTM). Then there is a change in the weights of all neurons in the neighborhood, but
this change depends on the distance between the winner and the neuron in the
neighborhood. In this case, the network is called a self-organizing map [4; 8].
3 Results
The authors collected statistical data on the yields of three main agricultural crops
(spring wheat, winter wheat, spring barley) for 30 years in the context of the
municipal districts of four regions of the Volga region (Samara, Saratov, Penza and
Ulyanovsk regions). Based on the data obtained, self-organizing Kohonen maps were
constructed for each culture separately. Maps were built according to data from 110
municipalities. The size of the map was chosen 7 by 8, based on the experiments
carried out, this size is optimal for such a number of objects. Figures 2-4 show maps
of three crops.
Output neurons
Input neurons
Fig. 1. The structure of the neural network of a self-organizing Kohonen map.
Fig. 2. Kohonen's map of spring barley.
Fig. 3. Kohonen's map for spring wheat.
Fig. 4. Kohonen's map for winter wheat.
As a result of building Maps, three clusters (with high, low and medium
productivity) were obtained for each crop. More details are presented in Table 1.
Table 1. Distribution of districts by clusters based on the results of constructing Kohonen
maps.
The number of districts of the region
Culture Cluster Samara Ulyanovsk
Penza region Saratov region
Region region
tall 17 2 1 9
Spring barley middle 8 13 8 5
low 1 12 28 6
tall 23 1 – 6
Spring wheat middle 2 17 12 8
low 1 9 25 6
tall 17 7 3 6
Winter wheat middle 6 16 12 8
low 3 4 22 6
As a result, municipalities were singled out that fell into the same cluster for three
cultures at once. So, 17 municipalities got into the "high" cluster (13 from the Penza
region, 3 from the Ulyanovsk region and one from the Samara region), 27
municipalities got into a cluster with low productivity (2 from the Samara region, 20
from the Saratov region and 5 from the Ulyanovsk region). oblast) and 66
municipalities fell into the cluster with an average yield (13 from the Penza region, 24
from the Samara region, 12 from the Ulyanovsk region, 17 from the Saratov region).
4 Discussion
The results obtained can be used by insurance companies and agricultural enterprises
to assess the prospects of using the agricultural insurance tool in their activities. If an
enterprise has been operating for some time, then it has accumulated experience, it
knows the specifics and features of the territory in which it conducts economic
activity, but if the enterprise expands the boundaries of production or moves to other
regions or territories, then the lack of objective data will limit it in making
management decisions. Based on the data obtained, knowing the territory, it is
possible to assess the prospects and expected average results for this zone, and also to
determine which cluster the processed territory will belong to. A cluster with high
yields for three crops speaks of favorable agro-climatic conditions for growing the
studied crops and these territories are interesting for the insurance company, and a
lower tariff can be offered for them. For a cluster with low yields for three crops, we
can talk about not very favorable agro-climatic conditions for growing the studied
crops and for the insurance company these territories are risky, since they will
increase the losses of the insurance company, as a result, they will offer them a higher
tariff to ensure break even work. The cluster with an average yield, in turn, is not a
stable category, since it includes areas in which both low, high and medium yield are
observed, however, a more detailed analysis of the production structure will allow
choosing the optimal insurance rate.
5 Conclusion
Further research will focus on the inclusion of other crops. The construction of
Kohonen maps for new cultures will make it possible to more accurately identify
clusters, to determine the specialization of specific territories. Also, work in the future
will be aimed at building geographic information systems that take into account not
the subject of identity, but the similarity of agro-climatic, technological, technical and
other characteristics of agricultural production.
6 Acknowledgments
The study was carried out within the framework of the scholarship of the President of
the Russian Federation for young scientists and graduate students for 2021-2023.
References
1. Ergunova, O., Skuratov, A., Pozdeeva, O., Kindaev, A.Y. Opportunities and emerging
trends for digitalization of Russian economy. IOP Conference Series: Materials Science
and Engineering, 483(1), 012115 (2019).
2. Draper, P., Paudyal, K. Information asymmetry and Bidders' gains. Journal of Business
Finance and Accounting, 35(3-4), 376-405 (2008).
3. Leung, C.M., Kwok, Y.K.: Erratum to: Real options signaling game models for dynamic
acquisition under information asymmetry. Decisions in Economics and Finance, 41(1), 35-
63 (2018).
4. Markowska-Kaczmar, U., Ros`oł, K.: Kohonen’s Neural Network and Evolutionary
Algorithms in Searching for Financial Investment Strategy. Proceedings of the
International Multiconference on Computer Science and Information Technology, 75–80
(2009).
5. Moiseev, A., Kindaev, A.Yu.: Simulation of insurance risks in agriculture. Journal of
Applied Engineering Science, 13(4), 257–264 (2015).
6. Moiseev, A., Lizunkov, V., Kindaev, A.Yu.: Insurers and insured individuals interaction as
basis for persistent agricultural production process. European proceedings of social and
behavioural sciences, 26. 532-540 (2017).
7. Nugroho, A., Takahashi, M., Masaya, I.: Village fund asymmetric information in disaster
management: Evidence from village level in Banda Aceh City. IOP Conference Series:
Earth and Environmental Science, 630(1), 012011 (2021).
8. Oja S., Kaski E.: Kohonen Maps. Elsevier Science (2003).
9. Ranaldo, A.,Somogyi, F.: Asymmetric information risk in FX markets. Journal of
Financial Economics Financial Economics, 140(2), 391-411 (2021).
10. Schneider, M., Stephenson, D.G.: Bargains, price signaling, and efficiency in markets with
asymmetric information. Games and Economic Behavior, 128, 160-181 (2021).
11. Yan, J., Yu, W., Zhao, J.: How signaling and search costs affect information asymmetry in
P2P lending: the economics of big data. Financial Innovation, 1(1), 19 (2015).