ANALYSIS AND MODELING OF DYNAMICS OF REVENUES TO THE BUDGETS OF THE UNITED TERRITORIAL COMMUNITIES OF THE ZAPORIZHZHIA REGION

The United Territorial Community (UTC), as an object formed by the voluntary amalgamation of territorial communities within one region, has a single local self-government body endowed with additional powers and budget revenues. Currently, an urgent problem for each UTC is the awareness not only of the amount of budget revenues, but their structure, the nature of the dynamics as a whole and in terms of their types. The UTC revenue planning requires their forecasting, and the use of relevant forecasting tools necessitates the study of the peculiarities of their structure and dynamics. The purpose of this work is to analyze the structure and dynamics of revenues to the budgets of the UTC of Zaporizhzhia region and to develop recommendations to community leaders on the choice of object, models and methods of forecasting revenues. The analysis of the current state of decentralization reform in Zaporizhzhia region has been carried out in the work. The dynamics and structure of revenues to the budgets of the UTC of the Zaporizhzhia region for 2016-2019 have been analyzed. The tendencies of dynamics of formation of incomes of local budgets have been defined. A cluster analysis has been conducted for the budget of Veselivska UTC, on the basis of which it has been found that the revenues are divided into four groups that have similar dynamics. A fractal analysis of receipts has been performed. Based on the analysis, recommendations for the selection of the object, models and methods for revenue forecasting have been developed.

The United Territorial Community (UTC), as an object formed by the voluntary amalgamation of territorial communities within one region, has a single local selfgovernment body endowed with additional powers and budget revenues. Currently, an urgent problem for each UTC is the awareness not only of the amount of budget revenues, but their structure, the nature of the dynamics as a whole and in terms of their types. The UTC revenue planning requires their forecasting, and the use of relevant forecasting tools necessitates the study of the peculiarities of their structure and dynamics. The purpose of this work is to analyze the structure and dynamics of revenues to the budgets of the UTC of Zaporizhzhia region and to develop recommendations to community leaders on the choice of object, models and methods of forecasting revenues. The analysis of the current state of decentralization reform in Zaporizhzhia region has been carried out in the work. The dynamics and structure of revenues to the budgets of the UTC of the Zaporizhzhia region for 2016-2019 have been analyzed. The tendencies of dynamics of formation of incomes of local budgets have been defined. A cluster analysis has been conducted for the budget of Veselivska UTC, on the basis of which it has been found that the revenues are divided into four groups that have similar dynamics. A fractal analysis of receipts has been performed. Based on the analysis, recommendations for the selection of the object, models and methods for revenue forecasting have been developed.

________________________________________________________________________________________________
Statement of the problem State policy of Ukraine in the field of local selfgovernment is based on the interests of residents of territorial communities and provides for decentralization of power: the transfer of executive power to local governments a significant part of the powers and, importantly, resources. The UTC as an object formed by voluntary association of territorial communities within one region, has a single local government body, endowed with additional powers and budget revenues. Currently, an urgent problem for each UTC is the awareness not only of the amount of budget revenues, but their structure, the nature of the dynamics as a whole and in terms of their types. The UTC revenue planning requires their forecasting, and the use of relevant forecasting tools necessitates the study of the peculiarities of their structure and dynamics.
Analysis of recent studies and publications The research on the theoretical aspects of local budgets as the financial basis of the existence of local communities devoted to the works of many Ukrainian scientists, such as V. Andrushchenko [1], S. Bukovynskiy [2], A. Buryachenko [3], A. Kirilenko [4], O Kovalyuk [5], R. Khalkhash [6], M. Kulchytsky [7], V. Oparin [8], Y. Pasichnyk [9] and others. Despite significant scientific work in the field of local budget research, financial theory does not have a single common definition of the essence of local budgets, which requires further research to establish the economic essence of local budgets and their role in the current budget system of Ukraine. A number of national scientists have studied the problems of decentralization and the formation, development and reform of local self-government in Ukraine, among which are: O. Batanov [10], T. Baranovska [11], A. Buchynska [12], V. Gladiy [13], N. Kondratska [14], Yu. Pasichnyk [15], P. Lyubchenko [16], S. Seryogin [17], I. Yaroshenko [18], R. Grishko [19], T. Bardina [20] considers the problem of rural development in the context of reform. Among the scientific works devoted to the formation of local budget revenues and their balancing, it is necessary to single out the works of scientists: O. Kyrylenko [21], S. Yuri [22], V. Pogorelov [23]. The insufficient attention is given to the analysis of the structure and dynamics of revenues to the budgets of the UTC. Also remain unresolved following questions: determining the capacity of newly formed communities; assessment of economic development of communities; the impact of the UTC budget revenues on the investment attractiveness of communities.
Objectives of the article The objective of this work is to analyze the structure and dynamics of revenues to the budgets of the UTC in the Zaporizhzhia region and develop recommendations to community leaders on the choice of object, models and methods of forecasting revenues.
According to the goal, the tasks were set: to analyze the dynamics of revenues to the budgets of the UTC in the Zaporizhzhia region, its nature, identified trends in the dynamics of local budget revenues; conduct a cluster analysis of the dynamics and fractal analysis of receipts; develop recommendations for the selection of facilities, models and methods for revenue forecasting.
The main material of the research In the Zaporizhzhia region, decentralization reform is being actively implemented, which affects the level of financial capacity of administrative-territorial units, changes in relations between state and local budgets. As of January 2020, there are 61 UTCs operating in Zaporizhzhia region [24], which is 5.7% of UTC in Ukraine, of which: in 2015 6 UTCs were formed, since 2016 another 10 UTCs have been operating in the oblast, since 2017 -20 UTC, in 2018 -8 UTC, in 2019 -17 UTC. Thus, 75.67% of the area of the region and 31.0% of the population are covered by UTC, formed as a result of reforming the administrative-territorial system. The availability of expenditures related to the performance of their functions by local authorities determines the need for revenue. Planning of local budget revenues requires an in-depth analysis not only of the level achieved, but also the dynamics of budget revenues. Comparisons can be made both by types of revenues and by subjects of revenues (individuals or legal entities, budgets of lower levels, etc.). The main objectives of the analysis and modeling are the study of the state of budget revenues; forecasting (forecasting) future revenues; improving the management of revenue sources. Thus, analysis and forecasting are the basic components of UTC management technology, the basis for a more balanced formation of their budgets. The first step in analyzing the dynamics of time series is to visualize the behavior of time series by constructing appropriate graphs and charts. Figure 1 shows the dynamics of total revenues to the budgets of the first 6 UTCs of the Zaporizhzhia region for the period from January 2016 to September 2019. Based on the visual analysis of the graphs of Figure 1, it is possible to draw the following preliminary conclusions: − revenues in the budgets of each community show an increasing trend throughout the period under study; − significant seasonal fluctuations in revenues are not observed during the study period; − Veselivka UTC has from 1.8 to 2.6 times higher revenues than other UTCs, due to the excess of the population from 2 to 3.8 times compared to other communities and a large area, so for a practical analysis of aspects of local budget revenues this UTC has been chosen. In 2019, total revenues increased by 25.9% and amounted to UAH 113 million.
To analyze the dynamics of revenues to the budget of Veselivska UTC, the relevant time series were visualized by constructing graphs and charts. Figure 2 shows the dynamics of total revenues to the budgets of Veselivska UTC for the period from January 2016 to September 2019. -18000000 -"Local taxes"; -19000000 -"Other taxes and fees". Let's analyze the structure of tax revenues presented in Figure 3.  In different months, the share of different types of revenues changes, but the overall process is more or less stationary. First of all, PIT and local taxes and fees need to be analyzed, and it is also advisable to pay attention to domestic taxes on goods and services (code 14000000). Analysis and grouping of revenues by similarity of dynamics Let's perform a cluster analysis of revenues to the budget of Veselivska UTC using two methods: PAM method and the Ward's method [26,27]. The input data for the analysis are monthly data on revenues to the budget of Veselivska UTC for 2016-2019. For the cluster analysis, 62 indicators were used -the amount of tax revenues of Veselivska UTC for the period from 2016 to September 2019.
The study used a statistical environment R [28], which is designed to implement models of Data Mining -a modern direction of in-depth data analysis to identify hidden patterns [29]. As a result of application of the Ward's method, the dendrogram of similarity of taxes (Fig. 5) -a tree of associations of clusters with the specified taxes on a horizontal axis and a scale of distances on a vertical has been received.    6. Diagram of silhouettes when divided into 4 clusters One of the tasks of cluster analysis is to select the optimal value of the number of clusters . In Fig. 7 presents a graph from which we can make assumptions about the optimal number of clusters: from 4 to 10. Fig. 7. Selection of the optimal number of clusters by the Elbow method and GAP-statistics Based on the study, four clusters were identified by the PAM method. The PAM is a modification of the k-mean algorithm, the k-medoids algorithm. The algorithm is less sensitive to noise and data emissions than the k-means algorithm because the medoid is less susceptible to emissions. Instead of calculating centroids, k is searched for the most representative objects (or medoids) among the analyzed observations. The intracluster scatter is estimated from the Manhattan distance, not the Euclidean distance. The PAM is effective for small databases, but should not be used for large data sets. The results of cluster analysis are presented in Fig. 8. The first cluster includes the two largest items of incomepersonal income tax (the share in budget revenues is 20%) and the educational subvention from the state budget (21%). Together, they form up to 41% of UTC budget revenues.
The second cluster includes articles with an average impact on budget formation, but with a small share in the budget, together forming 20% of the budget. Among them are excise tax on fuel, land tax on individuals, rent from individuals, revenues from fees for services provided by budgetary institutions in accordance with the law. The third cluster includes items of revenue, which form 32% of the budget. Mainly due to the single tax on agricultural producers, the basic subsidy to the local budget and the medical subvention. The fourth cluster includes one item of revenues -a subvention from the state budget to local budg Fig.8. The schedule of distribution of articles of receipts to the budget on clusters The first cluster includes the two largest items of income -personal income tax (the share in budget revenues is 20%) and the educational subvention from the state budget (21%). Together, they form up to 41% of UTC budget revenues. The second cluster includes articles with an average impact on budget formation, but with a small share in the budget, together forming 20% of the budget. Among them are excise tax on fuel, land tax on individuals, rent from individuals, revenues from fees for services provided by budgetary institutions in accordance with the law. The third cluster includes items of revenue, which form 32% of the budget. Mainly due to the single tax on agricultural producers, the basic subsidy to the local budget and the medical subvention. The fourth cluster includes one item of revenues -a subvention from the state budget to local budgets for the formation of the UTC infrastructure, which is 7% of budget revenues.
Analysis of revenue predictability The next step is to analyze the depth of memory of time series of budget revenues based on fractal analysis methods. As a result of the analysis, Hurst exponents were determined [30] for selected monthly revenues to Veselivska UTC, which are presented in Table 2. The presence of the property of persistence (trend resistance) was detected only in the time series of revenues under the code 11010100 (PIT). This means that the time series of such receipts has a long-term memory, and to predict if it is necessary to use methods that take this into account. Other time series of revenues are likely to be random, and it is sufficient to use statistical forecasting methods to predict them. We will conduct a comprehensive fractal analysis for the time series of receipts under the code 11010100. In Fig. 9 presents an example of H-trajectory and R / S-trajectory for the time series of receipts code 11010100. Fig. 9. Example of H-trajectory and R \ S-trajectory for the time series of receipts code 11010100 Fig. 9 shows that the breakdown of the trend on the R \ S -trajectory occurs at the value (on the seventh element of the trajectory-breakdown, the points on the graph are displayed starting from the third).
In Fig. 10 constructs a fuzzy set of memory depth of the time series of receipts under the code 11010100. From Figure 10 we see that the depth of memory, which is most common, is the value -it corresponds to half a year (6 months). However, the center of gravity of the fuzzy set is the value of l = 8.6 9, i.e. 3 quarters. These parameters can be used to determine the parameters of forecasting models.. The calculation of the Hurst exponent showed that the presence of the property of persistence (trend resistance) was detected only in the time series of revenues under the code 11010100 (PIT). This means that methods that take this into account must be used to predict it. Other time series of revenues are likely to be random, and it is sufficient to use statistical forecasting methods to predict them. The results of complex factor analysis can be used to build forecasting models.
Conclusions The analysis of the dynamics and structure of revenues to the budgets of the united territorial communities of the Zaporizhzhia region for 2016-2019 allowed us to conclude that the revenues to the budgets of UTC depend entirely on two factors: favorable geographical location; level of economic development of the community -the location of business structures on its territory.
The analysis of receipts to the budget of Veselivskaya UTC for 2016-2019 is carried out, the detailed analysis of dynamics and structure of own receipts of the specified UTC is carried out. In particular, cluster and fractal analysis were performed. The structure of own revenues of the budgets of the selected UTCs of the Zaporizhzhia region differs insignificantly. The excess of Veselivska UTC revenues compared to the other five UTCs, which were also established in 2015, is due to the large size of the community and the large population. According to the results of the analysis of the structure and dynamics of revenues to the budget of Veselivska UTC in the Zaporizhzhia region, the following conclusions can be reached: the priority task of the community is to create conditions for increasing the welfare of the population by activating the internal potential of the community, modernization of enterprises of the real sector of the economy, further development of acquired and established enterprises; creation of favorable conditions for the development of entrepreneurship, the functioning of institutions and institutions of the social sphere on the basis of innovative structural changes; strengthening investment activity and international cooperation; implementation of energy saving measures in the budget sphere; Veselivska UTC's own revenues are approximately equal to the proceeds from official transfers. The share of official transfers is 42.82%, which is one of the lowest rates among the other 5 oldest UTCs in the Zaporizhzhia region. This is due to a larger amount of own revenues, namely tax, the share of which is more than 50% in the structure of total community income; the structure of tax revenues is dominated by personal income tax, the share of which is 44% (is almost 25% of total income). That is, it can be argued that personal income tax is the main budget-generating revenue; the result of clustering was the division of all 62 items of income (own and transferred) into 4 clusters. The first group consists of the main budget-generating revenuespersonal income tax and education subvention, which together form up to 41% of budget revenues. The second group includes articles with a medium impact on budgeting. Each individual article forms no more than 1% of the total community income. The third group includes 7 items of revenue, which have an average impact on the formation of budget revenues due to its own share in total budget revenues (each is from 2 to 4). The fourth cluster includes one item of revenues -a subvention from the state budget to local budgets for the formation of UTC infrastructure, which is 7% of budget revenues and has a weak impact on the formation of UTC budget revenues; the calculation of the Hurst exponent showed that the presence of the property of persistence (trend resistance) was detected only in the time series of revenues under the code 11010100 (PIT). This means that it is necessary to use methods that take this into account to predict it. Other time series of revenues may be random. The results of complex factor analysis can be used to build forecasting models; analysis of the main statistics of time series of the following revenues: land tax on individuals, land tax on legal entities, single tax on agricultural producers, rent from individuals, personal income tax from business activities showed that the signs vary greatly and deviate significantly from the law of normal distribution.