Modeling of budget spending of Donetsk region
In current Ukrainian forecasting practice budget expenditures are often planned according to achieved results, taking into account the inflation rate. But this principle does not allow defining medium- and long-term trends, which provides evidence of lack of adequate forecasting of local budget expe...
Збережено в:
| Дата: | 2014 |
|---|---|
| Автори: | , |
| Формат: | Стаття |
| Мова: | English |
| Опубліковано: |
Інститут економіки промисловості НАН України
2014
|
| Назва видання: | Економіка промисловості |
| Теми: | |
| Онлайн доступ: | http://dspace.nbuv.gov.ua/handle/123456789/64024 |
| Теги: |
Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Цитувати: | Modeling of budget spending of Donetsk region / O.V. Sokolovska, D.B. Sokolovskyi // Економіка промисловості. — 2014. — № 1 (65). — С. 56-65. — Бібліогр.: 15 назв. — англ. |
Репозитарії
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-64024 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-640242014-06-10T03:01:26Z Modeling of budget spending of Donetsk region Sokolovska, O.V. Sokolovskyi, D.B. Macroeconomic and regional problems of industrial development In current Ukrainian forecasting practice budget expenditures are often planned according to achieved results, taking into account the inflation rate. But this principle does not allow defining medium- and long-term trends, which provides evidence of lack of adequate forecasting of local budget expenditures. Now the scientifically-based approach toforecast local budget expenditures is required; the latest is impossible without using of mathematical and economic models. Given paper is aimed to develop scientifically based methods and models in order to forecast local budget expenditures and to make a medium-term forecast of local budget expenditures for Donetsk region. In order to choice an appropriate forecasting model, based on existing theoretical issues, we distinguished three classes of forecasting models: microsimulation models, component-based models, regional-level models. This analysis allowed us to determine the forecasting technique which is the mixed variant of deterministic and econometric models. It based on using of correlatable factors, which influence directly on benchmark parameter - budget expenditures. Such technique provides the medium-term forecasting of budget expenditures in Donetsk region for 2014-2016. Input model data includes official statistical data for all considering indexes and also for benchmark parameter, covering period 2006- 2013, by half-year. After preliminary estimates we’ve chosen two forecasting models of multivariate regression type: additive and multiplicative (logarithmic) models. Modeling results showed that ratio between budget expenditures in Donets region and Ukraine’s GDP is sufficiently stable, it changes continuously according to political and economical government decisions; the saccadic changes can be naturally explained by hypothesis of external pulse effects (as it was in the second half of 2008). The chosen models allowed us to make a medium-term forecast of local budget expenditures of Donetsk region. Results of forecasting, as well as analytical conclusions can be useful for budget management in Donetsk region. Developed mathematical economic models can be used to forecast spending of local budgets of Ukraine. Розглянуто науково обґрунтовану пропозицію методів і моделей прогнозування видатків місцевих бюджетів. Серед інших обрано дві найбільш адекватні моделі багатовимірної регресії: адитивну та мультиплікативну. Порівняння результатів показало, що обидві моделі досить точно і приблизно однаково апроксимують вихідні дані. За їх допомоги здійснено прогноз бюджетних видатків на середньострокову перспективу для Донецької області. Рассмотрено научно обоснованное предложение методов и моделей прогнозирования расходов местных бюджетов. Среди прочих выбраны две наиболее адекватные модели многомерной регрессии: аддитивная и мультипликативная. Сопоставление результатов показало, что обе модели достаточно точно и примерно одинаково аппроксимируют выходные данные. С их помощью осуществлён прогноз бюджетных расходов на среднесрочную перспективу для Донецкой области. 2014 Article Modeling of budget spending of Donetsk region / O.V. Sokolovska, D.B. Sokolovskyi // Економіка промисловості. — 2014. — № 1 (65). — С. 56-65. — Бібліогр.: 15 назв. — англ. 1562-109Х http://dspace.nbuv.gov.ua/handle/123456789/64024 336.1:352(477.62) en Економіка промисловості Інститут економіки промисловості НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| topic |
Macroeconomic and regional problems of industrial development Macroeconomic and regional problems of industrial development |
| spellingShingle |
Macroeconomic and regional problems of industrial development Macroeconomic and regional problems of industrial development Sokolovska, O.V. Sokolovskyi, D.B. Modeling of budget spending of Donetsk region Економіка промисловості |
| description |
In current Ukrainian forecasting practice budget expenditures are often planned according to achieved results, taking into account the inflation rate. But this principle does not allow defining medium- and long-term trends, which provides evidence of lack of adequate forecasting of local budget expenditures. Now the scientifically-based approach toforecast local budget expenditures is required; the latest is impossible without using of mathematical and economic models. Given paper is aimed to develop scientifically based methods and models in order to forecast local budget expenditures and to make a medium-term forecast of local budget expenditures for Donetsk region. In order to choice an appropriate forecasting model, based on existing theoretical issues, we distinguished three classes of forecasting models: microsimulation models, component-based models, regional-level models. This analysis allowed us to determine the forecasting technique which is the mixed variant of deterministic and econometric models. It based on using of correlatable factors, which influence directly on benchmark parameter - budget expenditures. Such technique provides the medium-term forecasting of budget expenditures in Donetsk region for 2014-2016. Input model data includes official statistical data for all considering indexes and also for benchmark parameter, covering period 2006- 2013, by half-year. After preliminary estimates we’ve chosen two forecasting models of multivariate regression type: additive and multiplicative (logarithmic) models. Modeling results showed that ratio between budget expenditures in Donets region and Ukraine’s GDP is sufficiently stable, it changes continuously according to political and economical government decisions; the saccadic changes can be naturally explained by hypothesis of external pulse effects (as it was in the second half of 2008). The chosen models allowed us to make a medium-term forecast of local budget expenditures of Donetsk region. Results of forecasting, as well as analytical conclusions can be useful for budget management in Donetsk region. Developed mathematical economic models can be used to forecast spending of local budgets of Ukraine. |
| format |
Article |
| author |
Sokolovska, O.V. Sokolovskyi, D.B. |
| author_facet |
Sokolovska, O.V. Sokolovskyi, D.B. |
| author_sort |
Sokolovska, O.V. |
| title |
Modeling of budget spending of Donetsk region |
| title_short |
Modeling of budget spending of Donetsk region |
| title_full |
Modeling of budget spending of Donetsk region |
| title_fullStr |
Modeling of budget spending of Donetsk region |
| title_full_unstemmed |
Modeling of budget spending of Donetsk region |
| title_sort |
modeling of budget spending of donetsk region |
| publisher |
Інститут економіки промисловості НАН України |
| publishDate |
2014 |
| topic_facet |
Macroeconomic and regional problems of industrial development |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/64024 |
| citation_txt |
Modeling of budget spending of Donetsk region / O.V. Sokolovska, D.B. Sokolovskyi // Економіка промисловості. — 2014. — № 1 (65). — С. 56-65. — Бібліогр.: 15 назв. — англ. |
| series |
Економіка промисловості |
| work_keys_str_mv |
AT sokolovskaov modelingofbudgetspendingofdonetskregion AT sokolovskyidb modelingofbudgetspendingofdonetskregion |
| first_indexed |
2025-07-05T14:46:32Z |
| last_indexed |
2025-07-05T14:46:32Z |
| _version_ |
1836818676646412288 |
| fulltext |
–––––––––––––––––––––––––– Економіка промисловості Economy of Industry ––––––––––––––––––––––––––
56 ISSN 1562-109X
2014, № 1 (65)
UDC 336.1:352(477.62) Olena Vasilivna Sokolovska,
PhD in Economics,
Dmytro Borysovych Sokolovskyi,
PhD in Economics
The Institute of the Economy of Industry
of the NAS of Ukraine, Donetsk
MODELING OF BUDGET SPENDING OF DONETSK REGION
Current Ukrainian economic development
is characterized by problems in tax and expendi-
ture policy caused by imperfection of budgetary
laws which regulates organizational and finan-
cial relations within public finance. The success-
ful administration of public finance apart from
everything else means the creating conditions
for effective industrial development.
The planning and forecasting of local
budget expenditures is an integral part of such
administration. In domestic practice the devel-
opment of scientifically-based techniques of
forecasting of local budget expenditures using
the mathematical economic models will contri-
bute to effective regulation of social and eco-
nomic processes in public finance area.
In practice, the process of budget fore-
casting is an element of a budget planning.
However, there are some differences in this
field. Firstly, the planning is a kind of strategic
prediction of performance of the local govern-
ment and the community, whereas the forecast-
ing is a revision of the local budget to reflect
changing market conditions [13, p. 410].
C. Swanson noted that an effective forecast
model presents a range of possible outcomes,
based on a set of diagnosed variables and as-
sumptions [14, p. 60]. A. Harvey explained that
forecasts of the budget categories are made by
extrapolating the components estimated at the
end of the sample [10, p. 14].
But, according to P. Galinski, the accura-
cy of this method is especially tied with the eco-
nomic, legal and political stability, both in the
country and the community. However, using
only time-series forecasts in local governments
during the budget preparation may cause some
negative consequences, such as:
the constraint of the activity of the local
authorities in the field of anticipating the poten-
tial future events which determine the local
budget;
repeating the wrong decisions from the
past;
the appearance of uneconomical opera-
tions in local authorities;
preparing the future budget taking advan-
tage of the data from the actual budget, which is
still the kind of the plan [9, p. 219-220].
Moreover, the EU practices showed that
the Medium-Term Expenditure Framework
(MTEF), based on simple extrapolation of ex-
penditure and revenue appropriations, assuming
similar trends set by the current policies, and
introduced in several countries of South Eastern
Europe was not very successful, because there
are rather frequent changes in local government
funding schemes, which cannot be predicted by
the MTEF [12, p.7].
To avoid those negative consequences,
the bulk of issues, related to budget forecasting,
includes other econometric techniques for plan-
ning the budget revenues and spending at local
level.
Thus, the experts of World Bank ex-
plained the expenditure forecasts on local levels
for different parameters: personnel expenditures,
non-personnel current Expenditures, direct pay-
ments to special-needs households, capital ex-
penditures and debt service. They said that
“Multiyear projections of spending are general-
ly built from accounting identity models under
specific assumptions regarding levels of service”
[11, p. 69], however they noted that “Different
techniques can be used to forecast both revenues
and expenditures. They range from simple
judgmental approaches that rely on the know-
ledge of experts to sophisticated multivariate
statistical techniques” [11, p. 74]. The World
Bank experts distinguish following general fore-
casting techniques, which can be more applica-
ble to forecast local revenues and expenditures.
Judgmental Techniques
Judgmental forecasting essentially relies
on the forecaster’s special expertise – that is,
knowledge of the local revenue system and the
factors that tend to affect annual flows of reve-
nue. Because this subjective approach is pri-
© O.V. Sokolovska, D.B. Sokolovskyi, 2014
–––––––––––––––––––––– Економіка промисловості Экономика промышленности ––––––––––––––––––––––
ISSN 1562-109X 57
2014, № 1 (65)
marily dependent on the idiosyncrasies of the
specific situation and forecaster, not much can
be said about it other than that its implementa-
tion cost is likely to be low and that it can yield
fairly accurate short-term forecasts.
Time-Series Techniques
Time-series techniques link expected fu-
ture revenues or expenditures to past experience.
These techniques can differ greatly in terms of
complexity. Trend techniques are simple to use
and to explain, but they rest on the assumption
that the factors that have influenced a revenue or
expenditure in the past will continue to exist.
Deterministic Techniques
Forecasters may find variables other than
the passage of time more realistic as determi-
nants of future revenues or expenditures. Fore-
casters use deterministic forecasts extensively in
making projections of expenditures. Determinis-
tic approaches to forecasting are quite simple.
Unlike time-trend techniques, they do not re-
quire that the forecaster assume that future reve-
nues or expenditures will rise (or fall) inexora-
bly as they have in the recent past. The tech-
nique does, however, require that the forecaster
make explicit assumptions regarding the varia-
ble(s) thought to drive the revenue or expendi-
ture being forecasted. Such assumptions may
turn out to be erroneous.
Statistical Models
Statistical forecasting models, sometimes
termed econometric models, constitute the most
complex approach to forecasting and require the
most extensive amount of data. They allow the
forecaster to attempt to capture the effects of
one or more variables that conceptually should
affect a revenue or expenditure and to base the
relationship between those variables and the one
being forecasted on statistical estimation tech-
niques. Because local economic conditions are
likely to affect local government revenues, reve-
nue forecasts from statistical modeling are more
common than spending forecasts from such
modeling. The accuracy of forecasts from this
technique relies on selection of reasonable inde-
pendent variables, the correctness of the pro-
jected values of those variables, and the stability
of the statistical relationship into the future.
Unlike judgmental techniques, the method
makes explicit the factors that the forecaster is
using to generate forecasts and therefore permits
ex-post analysis of erroneous forecasts so that
future forecasts might be improved. Unlike pro-
jections from trend-based forecasts, projections
from a statistical model will depend on the ex-
pected changes in one or more independent va-
riables; hence, the revenue or expenditure series
may show decreases as well as increases into the
future. Unlike the deterministic approach, the
statistical technique permits the analyst to learn
whether the hypothesized relationships between
the chosen independent variables and the reve-
nue/expenditure series are statistically relevant
(statistically significant) [11, p. 54-57].
In USA, notably in New-York state, to
forecast different local expenditures one can use
different techniques. Thus, medicaid forecast
provides a point-in-time estimate for program
spending based on an analysis of current and
historical claims and a number of other known
factors. These estimates can be subject to consi-
derable variance and are highly sensitive to eco-
nomic conditions. The welfare program forecast
methodology includes welfare caseload equa-
tions. Caseloads are estimated to vary based on
factors such as entry-level employment levels
and the State’s minimum wage. The models also
contain measures that attempt to capture the im-
pact of administrative and programmatic efforts
at the national, State, and local levels to reduce
welfare dependency. debt service forecast me-
thodology involves a multi-faceted approach to
forecast debt service costs. This includes fore-
casts for both fixed and variable interest rate
costs and projections for the amount of new
fixed and variable rate debt that is planned to be
issued to finance capital projects over the next
five year period [8, p. 181-226].
Ukrainian economists, in contrast, pay in-
sufficient attention to forecasting local spending
and revenues. I. Lukyanenko et al. [3] developed
a set of econometric models in order to forecast
different local revenues and expenditures. B.
Sylenkov [4] proposed a forecasting model of
local budget expenditures based on program-
oriented and goal-oriented approach. S. Legkos-
tup and G. Sukrusheva [2; 5] analyzed the fore-
casting techniques based on mathematic eco-
nomic models. I. Chugunov proposed the me-
thodology of forecasting of revenue part of local
budgets [6]. V. Vishnevskyy et al. [1] developed
a system of monitoring of local budgets which
involves the forecasting of local revenues and
spending.
In current Ukrainian forecasting practice
budget expenditures are often planned according
to achieved results, taking into account the infla-
tion rate. But this principle does not allow defin-
–––––––––––––––––––––––––– Економіка промисловості Economy of Industry ––––––––––––––––––––––––––
58 ISSN 1562-109X
2014, № 1 (65)
ing medium- and long term trends, which pro-
vides evidence of lack of adequate forecasting of
local budget expenditures. Now the scientifical-
ly-based approach to forecast local budget ex-
penditures is required; the latest is impossible
without using of mathematical and economic
models.
So, this paper is aimed to develop scien-
tifically based methods and models in order to
forecast local budget expenditures and to make a
medium-term forecast of local budget expendi-
tures for Donetsk region.
OECD’s experts in their work “A Com-
parative Analysis of Health Forecasting Me-
thods” analyzed the classification of mathemati-
cal economic models, used to forecast health
expenditures. Adjusting this analysis to local
budget expenditures, we distinguished the fol-
lowing classes of forecasting models.
Forecasting models typically project local
budget expenditure at the level of individuals,
groups of individuals or the community as a
whole. At the same time, models can focus on
specific sections of expenditure, such as health,
housing and community amenities, education
etc. By considering both the level of aggregation
of the units analyzed and the level of detail of
budget expenditure to be projected, it is useful to
identify three broad categories of budget ex-
penditure forecasting models (Fig. 1).
Fig. 1. Classes of local spending forecasting models
Models focusing on individuals as the unit
of analysis for the projection are referred to as
micro models. All examples of micro models in
this review use microsimulation techniques.
Those stratifying sections of budget expenditure
into groups, or stratifying individuals into
groups, or combinations of these two dimen-
sions, are identified here as component-based
models. Finally, macro-level models focus on
total expenditure as the unit of analysis. Within
this group, some regional-level models (for ex-
ample, computable general equilibrium models,
constructed on regional level) project future lo-
cal budget expenditure trends within the context
of the whole economy.
Microsimulation models
Microsimulation models are powerful
tools which allow analysis and testing of rela-
tively detailed “what-if” scenarios resulting
from a variety of policy options. The scenarios
can be very informative for policy makers as
they may provide information beyond what is
available from retrospective population studies.
The units of analysis of the microsimulation
–––––––––––––––––––––– Економіка промисловості Экономика промышленности ––––––––––––––––––––––
ISSN 1562-109X 59
2014, № 1 (65)
models are individuals. These individuals can be
aggregated into policy-relevant groups and ana-
lysed using relevant indicators such as inequali-
ty and poverty indices.
Microsimulation models reproduce the
characteristics and behaviour of a large sample
of individuals representing the whole population
of interest. To test the potential impact of a new
policy, the microsimulation model is run twice –
once with the base case or status quo scenario
and then again with a policy change or variant
scenario perturbing the environment in which
the individuals operate. These scenarios produce
a chain reaction where individuals react to the
policy changes first and then, depending on the
design of the model, may also react to the reac-
tion of other individuals.
The results are the potential future out-
come of the reform and are often compared with
the base case to evaluate the potential impact of
the reform. Microsimulation models require
large amounts of data to effectively assemble a
sample that adequately represents the whole
population of interest and includes all of the
characteristics of interest. Data are often ga-
thered from a variety of sources, and sophisti-
cated statistical techniques are often required to
standardize the various databases so that they
can be used to populate all of the desired
attributes of individuals included in the sample.
Component-based-models
The most widely used class of models is
component-based-models. This class includes a
large variety of forecasting models that analyse
budget expenditure by financing agents, by pro-
viders, by goods and services consumed, by
groups of individuals or by some combination of
these groups. When expenditures are grouped by
financing agents, the models often consist of
different layers, each of which may use a differ-
ent technique to project a sub-component of ex-
penditure. An important sub-class of compo-
nent-based models is represented by cohort-
based models. In cohort-based models, individu-
als are grouped into cells according to several
key attributes. Further refinements are obtained
by sub-dividing the cohorts according to other
commonly-used attributes.
These models are often referred to as ac-
tuarial models or cell-based models, where the
term cell identifies the sub-categories into which
each cohort is divided. Each cell in the model is
associated with an average cost of public goods
and services (usually expressed in real terms).
Future health expenditure is determined by mul-
tiplying the average costs by the projected num-
ber of individuals included in each cell. Cohort-
based models have been very common over the
years, probably because they offer a number of
advantages. First, their implementation and
maintenance tends to be simple and relatively
inexpensive. This is because this class of models
can be developed in an interactive spreadsheet,
requiring a limited amount of data and generally
including only a few parameters. Many of these
parameters can be found in the literature, rather
than being estimated. Secondly, the impact of
policy changes can be assessed easily by simply
modifying the policy parameters. Component-
based models are typically less data demanding
then microsimulation models which partially
explains their popularity. However, the devel-
opment of more sophisticated versions of the
component-based models could require addi-
tional information.
Regional-level models
Regional-level models restrict the analysis
to local budget expenditures. They are most ap-
propriate for short-term projections in the pres-
ence of clear and undisturbed trends and in the
absence of structural breaks. Therefore, these
extrapolation methods can benefit from the iner-
tia in the financial systems in the short-run.
Econometric regression analysis is used to
fit time-series data. Projections can be based on
pure extrapolation of the statistical models fit-
ting the data or they can be based on the pro-
jected values of the critical explanatory va-
riables, whenever included. The accuracy of
forecasts was then compared to the results ob-
tained from three different pure extrapolation
methods (exponential smoothing, moving aver-
age and ARIMA methods). Within the class of
regional-level models are “computable general
equilibrium (CGE)” models on regional level.
These are models that allow for the mea-
surement of broader consequences to the econ-
omy resulting from budget spending growth and
for feedback or reaction from individuals and
companies.
Regional-level models are typically the
least demanding in terms of data requirements.
This is particularly the case for pure extrapola-
tion methods which use a single time series and
for regression-based models which very often
include just a few explanatory variables. The
–––––––––––––––––––––––––– Економіка промисловості Economy of Industry ––––––––––––––––––––––––––
60 ISSN 1562-109X
2014, № 1 (65)
computational and data requirements for Dy-
namic Computable General Equilibrium Models,
on the other hand, are generally much higher
and depend on the specification of the equations
included in the model [7, p.18-22].
The aforesaid analysis allowed us to de-
termine the forecasting technique which is the
mixed variant of deterministic and econometric
models. It based on using of correlatable factors,
which influence directly on benchmark parame-
ter – budget expenditures. This technique pro-
vides the forecasting of budget expenditures in
Donetsk region for 2014-2016. The volatility of
economic situation causes the reasonability for
medium and short term forecasting. We should
note that such forecasting is conceptually possi-
ble since the local budgets are inertial whereas
they are related to financial of social expendi-
tures, which can be sharply modified, so they
are sufficiently predictable.
Developing and parameterization of mod-
el of budget expenditures in Donetsk region
Ukraine as the rest transformation coun-
tries inherited some problems related to forecast-
ing the local budget expenditures, particularly:
lack of statistical data, caused by both of
their inaccessibility and sharp changes in social
and economic state policies, and also by strong
propensity for spillover externalities, which
even in presence of large arrays of economic
data, makes them less informative and allows
using only up-to-date information;
large relative share of inter-budgetary
transfers in local budget; the disproportionality
of formers distorts essentially the conceptual
logics of expenditures and complicates the fore-
casting of budget spending based on classical
techniques which evaluate total expenditures on
account of standard sectional expenditures;
lack of control on budget revenue and
spending on local level, which involves to take
into account the “contingencies” in forecasting
process; at this time such contingencies are hard
to account since they are not considered in sets
of regional economic indicators.
All aforesaid does not provide the possi-
bility of using the deterministic model at this
stage; one should be limited by set of stochastic
(or even trend) models for some indicators and
to test forecast accuracy on current statistical
data.
Verification of forecasting models of lo-
cal budget expenditures in Donetsk region
Model assumptions
Concerning the forecasting model of local
budget expenditures, we assume that:
it can be classified as stochastic multi-
dimensional model (additive, multiplicative or
transcendental logarithmic one);
model of budget expenditures at national
level conceptually is similar to model of budget
expenditures at local level;
parameters which influence on amount of
local budget expenditures, are the following:
global economic indices, macroeconomic index-
es and regional economic indexes (economic
indexes at regional level).
Independent variables
As independent variables, according to
the last assumption, we defined the following
indexes:
global economic indices – world energy
prices [15], particularly:
average oil price: Brent (Great Britain),
West-Texas Intermediate (USA) and PEC
Reference Basket of Crudes (xh1);
steam coal price (Australia) (xh2);
Russian natural gas border price in Ger-
many (xh3);
macroeconomic indexes:
official exchange rate (UAH vs USD)
(xm1);
inflation rate (on an accrual basis) (xm2);
GDP (xh3);
regional economic indexes:
population size in region (xl1);
average wage in region (xl2).
The choice of indexes is determined by
their direct influence on local budget expendi-
tures: buildings maintaining needs energy ex-
penditure; population size is directly related to
amount of social expenditure; also, wages rate
influences directly on amounts of inter-
budgetary transfers.
Statistical information
Input model data includes official statistic-
al data for all considering indexes and also for
benchmark parameter – local budget expendi-
tures, covering period 2006-2013, by half-year
(Table 1). The forecast was made for Donetsk
region for period of 2014-2016 both dates inclu-
sive.
–––––––––––––––––––––– Економіка промисловості Экономика промышленности ––––––––––––––––––––––
ISSN 1562-109X 61
2014, № 1 (65)
–––––––––––––––––––––––––– Економіка промисловості Economy of Industry ––––––––––––––––––––––––––
62 ISSN 1562-109X
2014, № 1 (65)
Choice of model type
Since the forecast was made on the basis
of statistical samples, we’ve chosen the possible
stochastic models (more specifically – models of
multivariate regression type)
We’ve made a choice between four di-
lemmas:
linear or non-linear model;
additive or multiplicative model;
use of all independent variables or only
naturally independent ones;
use as variable the minimum wage in
country or average wage in region.
Consequently, four dilemmas resulted in
sixteen variants of models. During three stages
of comparison, we’ve omitted the following
groups:
average wage in region was statistically
more adequate than minimum wage in country;
modeling using all initial variables
showed more adequacy in comparison with na-
turally independent variables, which are weakly
correlated;
models which are represented as additive
and multiplicative polynomials (transcendental
logarithmic function), regardless of virtually
absolute approximation of real data, showed the
poor forecast accuracy, compared to simple ad-
ditive (multivariate linear function) and multip-
licative (linear logarithmic function).
As a result the final comparative verifica-
tion was made for two models:
additive model – model of multivariate regres-
sion type for eight initial variables:
8
1
hi
hi
a
x
i
y x
.
multiplicative (logarithmic) model – mod-
el of multivariate regression type for logarithms
of eight initial variables:
8
1
hi
hi
a
x
i
y x
.
Results of verification
For each of model we’ve built corres-
ponding regression equation by 16 points (2006-
2014 period). Predicted independent variables
are presented in Table 2 (information for fore-
casting was taken from open sources).
Table 3 presents comparison of modeling
results.
Table 2
Forecast values of independent variables for model of budget expenditure
Period
Oil price,
USD per
barrel
Natural gas
price, USD per
thousand cubic
feet
Steam
coal
price,
USD/t
Exchange
rate (UAH
vs USD)
Inflation
rate (on an
accrual
basis)
GDP,
USD
billion
Population
size, thou-
sands
Average
wage,
UAH
2014,1 107,35 406,38 83,00 8,500 1,005 90,113 4341322 3795
2014,2 105,70 406,88 88,00 9,000 1,008 92,198 4326230 4104
2015,1 103,85 398,04 89,00 9,200 1,010 94,283 4311138 4263
2015,2 102,00 389,19 90,00 9,500 1,013 96,368 4296046 4642
2016,1 101,35 387,42 90,50 9,700 1,015 98,453 4280955 4807
2016,2 100,70 385,65 91,00 9,800 1,018 100,538 4265863 5266
Both models sufficiently exactly approx-
imate initial statistical data. Thereat, the multip-
licative model appears more exact in control
forecasting for first half-year (Table 3), while
the additive model is more stable in long-term
forecasting (Table 4, columns 2, 3).
Such conclusion does not allow choosing
either model for forecasting; thereat, it’s ad-
visable to made forecast as interval within
predicted values of additive and multiplicative
models.
Analysis of dependence between budget
expenditures and GDP
It can be made a logical assumption that
the benchmark parameter – budget expenditures,
depends proportionally on GDP, i.e. this ratio is
constant. Forecasting of budget expenditures by
means of multivariate regression a priori con-
firms this hypothesis, but only in the case when
other variables do not mar up because of autore-
gression.
Comparison in fact both of real statistical
and forecasting data (Table 5, Fig. 2) shows that
the given hypothesis is completely plausible for
real economic indicators for 2006-2013.
––––––––––––––––––––––––––––
1 At this case we might choose between prices
for all energy products or only coal prices.
–––––––––––––––––––––– Економіка промисловості Экономика промышленности ––––––––––––––––––––––
ISSN 1562-109X 63
2014, № 1 (65)
Table 3
Comparing of forecasts made by additive and multiplicative models
Period Real budget expendi-
tures, UAH million Additive model Multiplicative model
2006, 1 2918,20 3129,77 2992,905
2006, 2 3441,72 3070,15 3339,779
2007, 1 3704,83 3454,70 3584,707
2007, 2 4544,61 4420,25 4488,189
2008, 1 5302,82 5685,98 5442,340
2008, 2 5639,41 5831,45 5813,456
2009, 1 4903,16 4594,89 4797,156
2009, 2 6144,56 7190,37 6755,466
2010, 1 6792,60 6868,72 6740,357
2010, 2 8256,27 7947,33 8080,356
2011, 1 9202,07 8819,93 8922,352
2011, 2 11311,77 10967,72 10895,395
2012, 1 10064,62 10035,28 9927,105
2012, 2 11649,20 11133,82 11184,619
2013, 1 9511,26 9916,03 9882,421
2013, 2 10860,64 11181,32 11244,678
Approximation error 0,0069 0,0084
Table 4
Comparing of forecasts of local budget expenditures for Donetsk region for the period until 2016
Period Additive model Multiplicative model
2014, 1 10914,87 11108,767
2014, 2 11711,32 12023,861
2015, 1 11880,75 12225,008
2015, 2 12514,41 12862,257
2016, 1 12754,72 13094,455
2016, 2 13243,74 13454,115
Table 5
Share of budget expenditures of Donetsk region in GDP according
to model of multivariate linear regression
Year GDP of Ukraine, GDP,
USD billion
Donetsk region
Budget expenditures, GDP,
USD billion
Share of budget expenditures
in GDP
2006 511,392 6,360 0,0124
2007 672,717 8,249 0,0123
2008 939,976 10,942 0,0116
2009 867,540 11,048 0,0127
2010 1032,045 15,049 0,0146
2011 1249,544 20,514 0,0164
2012 1379,537 21,714 0,0157
2013 1398,020 22,875 0,0164
2014 1595,749 25,050 0,0157
2015 1782,903 27,071 0,0152
2016 1940,266 29,187 0,0150
–––––––––––––––––––––––––– Економіка промисловості Economy of Industry ––––––––––––––––––––––––––
64 ISSN 1562-109X
2014, № 1 (65)
Fig. 2. Comparative dynamics of budget spending of Donetsk region
and its share in Ukraine’s GDP
Thus, in 2006-2009 the share of budget
expenditures of Donetsk region in Ukraine’s
GDP stably reached the 1,23-1,27% and only in
crisis year of 2008 it declined to 1,16%. Hereaf-
ter, most likely because of political factors, the
value of this parameter has been increased to
1,46% in 2010; in 2011-2013 it ranged from
1,57-1,64%. According to forecast (made by
means of multivariate linear regression), starting
from 2014 this share should decrease nearly to
1,5% in 2016.
Such observations show on the one side
the plausibility of developed model concerning
its accordance to natural expectations and on the
other side – the tendency of state policy in area
of distribution of state funds.
Conclusions
According to findings we developed the
theoretical and methodological basis of forecast-
ing of local budget expenditures, particularly we
explained the expediency of use the econometric
methods and models, based on correlative fac-
tors, influencing directly in benchmark parame-
ter – budget expenditures.
In order to forecast local budget expendi-
tures we developed some mathematical econom-
ic models. Their further analysis allowed choos-
ing two models satisfying in the best way to re-
search goals: the additive model – the model of
multivariate linear regression for initial data and
the multiplicative (logarithmic) model for loga-
rithms of initial data.
Further we made verification for those
two models. The comparison of modeling results
showed that both models sufficiently exactly
approximated initial statistical data. At this, the
multiplicative model occurred more exact for
short-term forecasting, while the additive one is
more stable at long-term forecasting.
This conclusion does not allow choosing
either model for forecasting; thereat, it’s advisa-
ble to made forecast as interval within predicted
values of additive and multiplicative models
Modeling results showed that ratio be-
tween budget expenditures in Donets region and
Ukraine’s GDP is sufficiently stable, it changes
continuously according to political and eco-
nomical government decisions; the saccadic
–––––––––––––––––––––– Економіка промисловості Экономика промышленности ––––––––––––––––––––––
ISSN 1562-109X 65
2014, № 1 (65)
changes can be naturally explained by hypothe-
sis of external pulse effects (as it was in the
second half of 2008). This confirms the adequa-
cy of proposed model of expenditure forecast.
According to the model results for me-
dium-term forecast in 2016 with the expected
exchange rate 9,8 UAH/USD and expected in-
flation rate 0,5%, the budget expenditures of
Donetsk region can be expected approximately
as 26 UAH billion, that will be 1,5% of GDP.
Forecast values are defined at the current prices.
It should be noted that modeling results
should be defined more exactly in further, whe-
reas only by means of broaden economic analy-
sis one can improve accuracy of forecasting of
local budget expenditure in Ukraine.
Results of forecasting, as well as analyti-
cal conclusions can be useful for budget man-
agement in Donetsk region. Developed mathe-
matical economic models can be used to forecast
spending of local budgets of Ukraine.
References
1. Как обосновать бюджетно-нало-
говую политику государства? Опыт научного
проектирования и реализации автоматизиро-
ванной системы сопровождения бюджетного
процесса на региональном уровне: моногр. /
В.П. Вишневский, Р.Н. Лепа, А.В. Половян и
др.; под общ. ред. В.П. Вишневского / НАН
Украины, Ин-т экономики пром-сти. – До-
нецк, 2011. – 116 с.
2. Легкоступ І.І. Теоретичні та прак-
тичні аспекти видатків місцевих бюджетів
України в сучасних умовах / І.І. Легкоступ //
Економіка. Фінанси. Право. – 2010. – № 2.–
С. 22-26.
3. Лук'яненко І.Г. Економетричні під-
ходи до аналізу фінансової програми міс-
цевих органів влади України / І.Г. Лук'янен-
ко, Ю.О. Городніченко, Л.І. Краснікова. – К.:
Вид. дім «KM Academia», 2000. – 121 с.
4. Сілєнков Б.В. Прогнозування дохо-
дів і видатків місцевих бюджетів / Б.В. Сі-
лєнков // Вест. Херсон. гос. техн. ун-та. –
2003. – № 2 (18). – С. 7-9.
5. Сукрушева Г.О. Необхідність пла-
нування видатків місцевих бюджетів в сучас-
них умовах розвитку економіки України /
Г.О. Сукрушева // Вісник економіки транс-
порту і промисловості. – 2011. – № 36. –
С. 86-89.
6. Чугунов І.Я. Бюджетна система як
інструмент регулювання економічного роз-
витку: автореф. дис. ... д-ра екон. наук:
08.04.01 / І.Я. Чугунов ; НАН України, Ін-т
екон. прогнозування. – К., 2003. – 37 с.
7. Astolfi R. Comparative Analysis of
Health Forecasting Methods / R. Astolfi, L. Lo-
renzoni, J. Oderkirk // OECD Health Working
Papers, N59, OECD Publishing, 2012. – 121 p.
8. Cuomo A.M. Economic, Revenue,
and Spending Methodologies / A.M. Cuomo,
R.L. Megna. – New York State, 2013. – 228 p.
9. Galinski P. The accuracy of the budget
forecasting in local governments in Poland /
P. Galinski // Economics and Management. –
2013. – №18 (2). – P. 218-225.
10. Harvey A.C. Forecasting, structural
time series models and the Kalman filter / A.C.
Harvey. – Cambridge University Press, 2003. –
572 p.
11. Local Budgeting. Ed. by A. Shah. –
The International Bank for Reconstruction and
Development / The World Bank, 2007. – 410 p.
12. Péteri G. Local government budget-
ing practices: European practices and lessons for
Serbia / G. Peteri. – Council of Europe Office. –
Belgrade, 2009. – 15 p.
13. Sandu D.I. Multidimensional model
for the master budget / D.I. Sandi // Journal
of Applied Quantitative Methods. – 2009. –
Vol. 4. – P. 408-421.
14. Swanson C.J. Long-term financial
forecasting for local governments / C.J. Swan-
son // Government Financial Review. – 2008. –
Vol. 24. – № 5. – P. 60-66.
15. World Bank Commodity Forecast
Price data [Electronic resource] / The World
Bank. – Mode of аccess: http://data.world-
bank.org/data-catalog/commodity-price-data.
Received on 29.01.2014
1_65_4_P52
1_65_4_P53
1_65_4_P54
1_65_4_P55
1_65_4_P56
1_65_4_P57
1_65_4_P58
1_65_4_P59
1_65_4_P60
1_65_4_P61
|