Modeling the dynamic patterns of banking and non-banking financial intermediaries’ performance

Nowadays, there are many preconditions and circumstances for conducting shadow schemes in the financial market. Therefore, the level of risk of participation of bank and non-bank financial intermediaries in such schemes is assessed as high. The lack of a practical methodology for assessing the development trajectory of financial intermediaries raises the question of the need for preventive control and quality modeling of their growth dynamics. The study aims to identify and formalize the patterns of development paths of banking and non-banking financial intermediaries based on the Harrington desirability function, which will be used to identify risk patterns as indicative patterns of financial intermediaries’ participation in shadow schemes. The sample includes 13 banking institutions, 3 credit unions, 3 pawnshops, 3 insurance companies, and 3 financial companies. The obtained results showed the relationship between the financial intermediary risk level in terms of its participation in shadow schemes and the phases of the economic cycle as a catalyst for the economic dynamics of the formal and informal economy. Thus, in 2012–2015, most financial intermediaries were in the zone of most significant risk, especially banks, characterized by economic, social, and political instability. Today, banks are in the group with a controlled level of risk of participation in scheme operations. Over the years analyzed, a stable neutral level of risk of participation in shadow schemes was inherent in most non-bank financial institutions. They were less sensitive than banks to the phases of the economic cycle.
AcknowledgmentAlina Bukhtiarova and Yevgeniya Mordan gratefully acknowledge financial support from the Ministry of Education and Science of Ukraine (0120U100473, 0121U100469).


INTRODUCTION
Long-term economic growth depends on the level of investment, which is due to the efficiency of the structure of financial intermediaries in the economic system. The development and improvement of intermediary activities in the financial system increase the efficiency of savings and investment processes, positively affecting economic growth.
The shadow economy and corruption are the main threats to sustainable economic development. Shadow schemes are implemented through the movement of illegal financial flows. Most illegal financial schemes are carried out with financial intermediaries, whose arsenal of technologies and financial capabilities is changing rapidly under the pressure of the development of fintech and digital financial services. Based on globalization, the development of digitalization, automation, high mobility of consumers of financial services, the emergence of a generation of digital people who are always on the Internet or social networks, new financial intermediaries have entered the financial services market (fintech companies, P2P aggregators, crowdfunding platforms, digital wallets, robot advisors, ecosystems of digital e-commerce platforms). Without being bound by the classical norms of banking regulation and supervision, as well as an active focus on modern digital technologies, including cloud computing, APIs, cryptography, machine learning, biometrics, big data analytics, blockchain, artificial intelligence, and Internet things, allows banking and non-banking financial intermediaries actively increase the volume of financial activities. At the same time, they are significantly increasing the risks of their participation in shadow schemes and tax evasion.
There are many schemes of illegal financial flows in which banking and non-banking financial intermediaries participate. These schemes are carried out to legalize illegally obtained income, withdraw capital from the country, evade taxes, withdraw cash illegally through fictitious enterprises or pay for non-existent goods and services. Detecting "scheme" transactions are not easy, but it is possible because several factors indicate the ability of financial intermediaries to participate in shadow schemes.
Thus, modeling the activities of banking and non-banking financial intermediaries will identify existing trends and develop strategies for further development.

LITERATURE REVIEW
Financial intermediation has recently been perceived as an essential supporting mechanism for economic growth. Much attention in the scientific literature is paid to studying the role of banks, credit unions, insurance companies, and other financial institutions.
In general, scientists focus on studying the impact of local and global crumbs on the activities of individual financial intermediaries and the financial market. Therefore, Kozmenko et al. (2016) offer bank patterns evaluation based on Kohonen's self-organizing maps to determine further directions of financial institution strategies advanced under the influence of a disaster within the economy. The study used some guidelines for modeling the activities of banking intermediaries developed by the authors. At the same time, Plastun et al. (2018) inspect competitiveness within the stock market at some point of the local crisis of 2013-2015. The consequences advocate that the contemporary degradation of the Ukrainian inventory market is closely associated with good-sized changes within the marketplace attention resulting from the local crisis.
Many scientists study the role of financial intermediaries in the shadow sector of the economy. For instance, Tiutiunyk and Humenna (2021) examine and establish the scientists' work to evaluate the chance of economic intermediaries' participation in shadow transactions. The consequences of evaluating clinical guides on these problems show diverse tactics for analyzing those issues. Significant variations within the functioning of different international locations' monetary, banking, coverage, and funding markets have caused the need to develop and put into effect their methodologies for assessing the threat of participation of economic intermediaries in shadow transactions on the national stage. Moreover, Ozgur (2021) focuses on how shadow banking, known until recently as fringe and parallel banking, has emerged as a principal detail for the USA monetary system. Using current and new shadow banking indices, the author uses distinctive Markov switching models to discover the position of shadow banking on financial institution lending cycle dynamics in the USA.
It should be noted that some researchers focus on non-bank financial intermediaries, and others only on the banking sector. Thus, they share these markets without considering the shared banking and non-banking intermediation market. On the one hand, Aramonte et al. (2021) look at structural shifts in intermediation and how non-bank financial intermediaries have shaped the requirement and financial markets' liquidity inventory. On the other hand, Santandrea et al. (2018) present the most effective enterprise version configuration for public intermediaries. Also, Martinez-Miera and Repullo (2019) analyze the effects of bank capital requirements on the structure and risk of a financial system where markets, regulated banks, and shadow banks coexist. Banks face moral hazard when screening entrepreneurs' projects, and they could choose whether they need regulation. Oliynyk et al. (2017) profoundly investigated the activity of mixed life insurance intermediaries.
The following works are devoted to modeling the activities of financial intermediaries based on various quantitative and qualitative assessments of their activities. Thus, Boda and Zimkova (2018) offer a measure of monetary intermediation attainment that solves conditions, while the ability of economic intermediaries, from a macroeconomic perspective, can usually be decreased to taking deposits and imparting loans. Ghasemi et al. (2020) developed a quantitative monetary dynamic stochastic general equilibrium version with economical intermediaries and proposed endogenously determined stability sheet constraints. Also, Yang and Chang (2020) use the quantile regression approach to observe the uneven impact of middleman economic improvement on the monetary increase in low-and high-income countries. A three-zone neoclassical growth version contains a consultant circle of relatives, manufacturing, and middleman economic areas. The equilibrium answers decide the variables hired within the empirical version. This usually indicates that international locations should no longer expand economic intermediaries indiscriminately in pursuit of financial growth, especially for low-income countries. At the same time, Islam and Shah (2012) use cointegration and error correction mechanisms to test for causal relationship between the improvement in non-bank economic intermediaries and in line with per capita financial growth in Malaysia over the period 1974-2004.
The authors endorse that non-bank monetary intermediaries and financial growth are cointegrated. Financial growth is used as a structured variable, but no more, while the opposite variables are handled as fixed variables. The result also suggests a unique lengthy-run causal strolling from non-bank monetary intermediaries to per capita financial growth, rather than the other way around.

AIM, DATA, AND METHODOLOGY
The study aims to identify and formalize the patterns of development paths of banking and non-banking financial intermediaries based on the Harrington desirability function, which will be used to identify risk patterns as indicative patterns of financial intermediaries' participation in shadow schemes, and to explore the possibilities of transition of financial intermediaries between patterns (risk, neutral, under control) and changes in the characteristics of the patterns themselves at different phases of the economic cycle and stages of the life cycle of a financial intermediary.
It is proposed to apply five stages of building a model to estimate the trajectories of financial intermediaries.
Stage 1. Defining the system of indicators based on which the cluster map is built. To build the model, 25 Ukrainian financial intermediaries were selected, which functioned during 2012-2020. To test the model, a sample of banks, credit unions, pawnshops, insurance companies, and financial companies was formed. Table 1 presents the list of financial intermediaries included in the model. Among the selected indicators, there are both absolute and relative indicators that can characterize the effectiveness of financial intermediaries ( Table 2). Equity, UAH thousand fk3 Liabilities, UAH thousand fk4 Net income from sales of products, UAH thousand fk5 Other operating income, UAH thousand fk6 Other financial income, UAH thousand fk7 Financial result before tax (profit) UAH thousand fk8 Net financial result (profit), UAH thousand Stage 2. Normalizing model input data.
The study proposes using the comparative approach to rationing indicators used in mathematical statistics.
It determines the maximum or minimum of the data using the MAX or MIN formulas in the MS Excel software and normalizes the next step. Accordingly, normalized values by formula were found out (6).  Table 3.
The period 2012-2021 was chosen for the analysis. The calculation of indicators for Pivdennyi Bank as of January 1, 2020 is shown in Table 4.
Next, the weight of the indicators is considered and the convolution is performed.
The calculation of the synthesizing function G for each group of indicators as of January 1, 2003 is shown in Table 5. Viscovery SOMine is based on the concept and algorithms of Kohonen's self-organizing maps pack- Return on assets (ROA), % b1 Total assets, UAH thousand b3 Loans and receivables, UAH thousand b6

RESULTS
At the model's output, a set of Kohonen maps was obtained for selected groups of indicators and the boundaries of division into clusters. Based on the colors of the representation, the distance between the elements of the samples can be described.  The model's input data will be synthesizing functions G for 9 reporting dates.
Map scales can also be used to determine cell values, compare and analyze them (Figure 1).
It should be noted that eight clusters were obtained as a result of data processing. The general Kohonen map is shown in Figure 2.
The belonging of the studied financial intermediaries to the created patterns is presented on the example of cluster C1 (Table 6).
Thus, the trajectories of financial intermediaries during 2012-2020 were formed to elucidate the results.
The Harrington desirability function scale was used to analyze each cluster's estimates ( Table 7).
The formation of cluster ranks is presented in Table 8.
Based on the results obtained, the clusters were ranked (Table 9).
It is proposed to divide the clusters into groups (Table 10) conditionally to assess the effectiveness of each financial intermediary, which was assigned to a particular cluster.
Thus, among the 25 surveyed financial intermediaries that, as of January 1, 2020, operated in the financial market of Ukraine, the crisis in recent years could be observed in:    Stage 5. Assessing the model adequacy. Two conditional financial intermediaries are introduced to the study population, with "good" and "bad" values of indicators to verify the adequacy of the model. The model's reaction will conclude the correctness of the model's reaction to diametrically different values of indicators.
Finally, a new Kohonen map was obtained (Figure 3).
As a result of the introduction of conditional financial intermediaries, eight clusters were obtained. The structure of indicator groups is shown in Figure 4.
The formation of cluster ranks of the studied financial intermediaries is presented in Table 11.
Thus, the cluster rating was made (Table 12).       To assess the effectiveness of an individual financial intermediary assigned to a particular cluster, the clusters were conditionally divided into groups (Table 13). In the upper left corner there is cluster C7, the indicators of which show the best financial reporting data, and in the upper right corner there is cluster C5, on the contrary, the worst. The membership of a financial intermediary in these clusters is presented in Tables 14 and 15.  The added simulated financial intermediaries show an adequate model response to different input data values based on the obtained results.
The financial intermediary with underestimated indicators added to the study ends up in the worst pattern. The financial intermediary with inflated indicators gets into the best pattern, indicating the high quality of the proposed model for evaluating the pattern dynamics of financial intermediaries.
There is a clear relationship between the risk level of financial intermediaries and the probability of their participation in shadow schemes and the phase of the economic cycle of the economy. Thus, trajectories of financial intermediaries within individual patterns were formed (

CONCLUSION
This paper proposes a methodological approach to build a model for estimating the development trajectories of banking and non-banking financial intermediaries based on a set of patterns. Constructed patterns determine the level of probability of financial intermediaries' participation in illegal schemes based on Harrington's desirability function and Kohonen's self-organizing maps. The model uses 37 indicators that characterize the state of a particular group of 25 financial intermediaries. According to the model, the interaction trajectories of financial intermediaries were built into 8 patterns formed based on Kohonen's self-organizing maps and cluster analysis. This approach allows tracking the transition of financial intermediaries between patterns (risk, neutral, controlled) and changes in the patterns' characteristics at different stages of the economic cycle. During the analyzed period, the neutral level of participation risk in shadow schemes was inherent in most non-bank financial intermediaries (except for two credit unions and one pawnshop). It should be noted that non-bank financial intermediaries are less sensitive than banks to the phases of the economic cycle. According to the model, the riskiest patterns include pawnshops and credit unions.
The results obtained can further become the basis for de-shadowing tools that will take into account the microeconomic nature of business models of interaction between financial intermediaries and provide a significant positive macroeconomic and social effect.