CC BY
39
УДК 330; 330.4; 339.7 Н.С. Айюбова
DOI: http://dx.doi.org/10.21686/2500-3925-2022-2-14-22
Бакинский Государственный Университет, Баку, Азербайджан
Эконометрический анализ и моделирование динамики развития платежного баланса в Азербайджане
Цель исследования. Исследование посвящено эконометрическо-му анализу и моделированию динамики развития платежного баланса Азербайджана, формированию математико-стати-стического тренда, способной дать перспективную оценку развития платежного баланса. В соответствии с целью были поставлены задачи выбора наилучшего состава объясняющих факторов для модели, с помощью характеристик и критериев корреляционного и регрессионного анализа, эконометрических тестов расчет оценок характера и тесноты связи между объясняющими факторами, зависимым и независимыми факторами, проверки стационарности ряда.
Материалы и методы. Использованы официальные статистические данные Государственного Комитета Статистики и Центрального Банка Азербайджана, научные труды и исследования ученых, специалистов, как азербайджанских, так и зарубежных, в областях экономики и математико-экономического моделирования. Для эмпирического анализа нестационарных временных рядов в работе применены статистические методы обработки информации, для проверки адекватности и тестирования многомерной модели использованы соответствующие критерии и современные эконометрические процедуры с учетом воздействия экзогенных факторов. Для расчетов использованы пакеты прикладных программ, таких как Excel и Eviews 8. Результаты. Создана многомерная регрессионная модель, позволяющая проводить экономико-статистический анализ динамики счета текущих операций платежного баланса; определены форма и направления функциональной зависимости между зависимыми и независимыми переменными, оценена изменчивость переменных, проанализированы результаты многомерного регрессионного анализа по эконометрическим методикам; измерены и интерпретированы количественные характеристики механизмов влияния объясняющих факторов на платежный баланс; в модели исследованы корреляционные зависимости для причинно-следственных зависимостей, выполнен тест Грейнджера и выявлены факторы, достоверно
объясняющие исход с высокими вероятностями на основе критерия Фишера; стационарность модели измерялась на основе теста Дики-Фуллера. При разностях первой и второй степени стационарность модели авторегрессии определялась на основе критерия Стьюдента путем изменения величины лага. В процессе моделирования изначально построенная модель, охватывающая 1995-2017-е годы с 5-ю факторами как, иностранные инвестиции, экспорт, импорт, курс маната, общие инвестиции, показала недостаточную адекватность, то есть не стационарность ряда текущего счета платежного баланса. Курс национальной валюты, который участвует в модели как объясняющий фактор, подверг значения зависимого ряда большим колебаниям, росту дисперсии в остатках, что создала не стационарность и которое можно объяснить деноминацией национальной валюты в 2006 году. В последующем шаге был исследован период охватывающий 2006-2017-е годы. Также в процессе исследования в модель были добавлены независимые факторы, как дефицит государственного бюджета и валютные резервы. В результате была построена многофакторная эконометрическая модель.
Заключение. Построенная авторегрессионная модель достаточно адекватна, демонстрирует стационарность для временного ряда зависимой переменной и может считаться пригодной для прогнозных значений текущего счета платежного баланса. Для выработки конкретных рекомендаций перспективного развития платежного баланса, полученные результаты исследования, обоснованные проведенным анализом динамики развития платежного баланса, дают возможность выявить реальные тенденции платежного баланса Азербайджана по текущему счету и определить его взаимозависимость с другими макроэкономическими переменными.
Ключевые слова: платежный баланс, текущий счет, экономе-трическая модель, описательная статистика, вариабельность, стационарность, тест Дики Фуллера, тест Грейнджера.
Natavan S. Ayyubova
Baku State Universitety, Baku, Azerbaijan
Econometric analysis and modeling of the dynamics of the balance of payments' development in Azerbaijan
Purpose of the study. The study is devoted to econometric analysis and modeling of the dynamics of the balance ofpayments' development of Azerbaijan, the formation of a mathematical and statistical trend that can give a perspective assessment of the development of the balance of payments. In accordance with the goal, the tasks of choosing the best composition of explanatory factors for the model were set, using the characteristics and criteria of correlation and regression analysis, econometric tests, calculating estimates of the nature and closeness of the relationship between the explanatory factors, dependent and independent factors, testing the stationarity of the series. Materials and methods. The official statistical data of the State Statistics Committee and the Central Bank of Azerbaijan, scientific
works and studies of scientists, specialists, both Azerbaijani and foreign, in the fields of economics, mathematical and economic modeling were used. For the empirical analysis of non-stationary time series, statistical methods of information processing are used in the work; to check the adequacy and test the multivariate model, the appropriate criteria and modern econometric procedures are used, taking into account the impact of exogenous factors. For calculations, application packages such as Excel and Eviews 8 were used. Results. A multivariate regression model has been created that makes it possible to conduct an economic and statistical analysis of the dynamics of the current account of the balance of payments; the form and directions of the functional relationship between dependent
and independent variables were determined, variability of variables was estimated, the results of multivariate regression analysis using econometric methods were analyzed; the quantitative characteristics of the mechanisms of influence of explanatory factors on the balance of payments were measured and interpreted; correlation dependencies for causal dependencies were investigated in the model, the Granger test was performed and factors were identified that reliably explain the outcome with high probabilities based on the Fisher criterion; the stationarity of the model was measured based on the Dickey-Fuller test. With differences of the first and second degree, the stationarity of the autoregressive model was determined based on the Student's criterion by changing the lag value.
In the process of modeling, the initially constructed model, covering the years 1995-2017 with five factors such as foreign investment, exports, imports, manat exchange rate, general investments, showed insufficient adequacy, that is, non-stationarity of the current account series of the balance of payments. The exchange rate of the national currency, which is involved in the model as an explanatory factor, subjected the values of the dependent series to large fluctuations, an increase in the variance in the residue, which created non-stationarity
and which can be explained by the denomination of the national currency in 2006. In the next step, the period covering 2006-2017 was examined. In addition, in the process of research, independent factors were added to the model, such as state budget deficit and foreign exchange reserves. As a result, a multifactorial econometric model was created.
Conclusion. The constructed autoregressive model is quite adequate, demonstrates stationarity for the time series of the dependent variable and can be considered suitable for predictive values of the current account of the balance of payments. To develop specific recommendations for the long-term development of the balance of payments, the results of the study, substantiated by the analysis of the dynamics of the development of the balance ofpayments, make it possible to identify real trends in the balance of payments of Azerbaijan on the current account and determine its interdependence with other macroeconomic variables.
Keywords: balance of payments, current account, econometric model, descriptive statistics, variability, stationarity, Dickey Fuller test, Granger test.
Introduction
The dynamic integration of Azerbaijan into the world economic system, accompanied by the activation of transboundary flows of capital, the characteristics of economic development of the country's trading partners, changes in prices on world markets and other external factors has an absolute effect to the processes in the sphere of economic activity of the country.
In order to conduct analytical and scenario-forecasting researches and make management decisions, it is important to take into account the interdependen-cies in the domestic economy, as well as to identify the factors that shape the country's foreign relations in a comprehensive and systematic manner[1,2,3,4,5,6]. It should be noted that modern modeling "Tools" of macroeco-nomic analysis and forecasting are less focused on a comprehensive consideration of the characteristics of foreign economic activity and the traditional system of macroeconomic indicators. The researches conducted by experts in this area focus on the key indicators of trade balance, the mechanisms of influence of exports and imports on the development of the national economy, the identification of threats that create high dependence of production on world markets con-juncture[7,8]. The issues such
as the systematic study of other aspects of foreign economic activity, as well as the establishment of econometric models, are rare.
The research on the regulation of the balance of payments in the late 1990s and early 2000s was a part of a general analysis of the transit economy. However, the recent researches ha focused on the impact of national exchange rates on the balance of payments in the context of already established market relations[9,10]. By evaluating the impact of exchange rate fluctuations on the components of the balance of payments through the researches conducted, a model was developed, the theoretical and methodological bases of exchange rates were ana-lyzed[11], methods for estimating devaluation expectations of financial market entities were developed taking into account the dynamics of the currency structure of investment assets, an economic mathematical models has been established to determine the dependence of exchange rates on inflation processes.
The dedication of the research to the topical issue is determined by the processes in the field of foreign economic activity, which are characterized by significant and growing impact on the development of the national economy and with the importance of taking them into account in assessing the country's economic development trends in the future.
The need for modeling of processes related to foreign economic activity, assessment of socio-economic results, development of adequate management decisions to prevent risks of the economic system forms a demand for methodological and modeling tools[12,13]of forecasting researches and expansion of analytical capabilities. Based on them, it is important to create the necessary opportunities to study and fully take into account the dependences between macro-economic indicators of development and parameters of foreign economic activity. Along with the trade balance, the parameters of foreign economic activity include indicators such as the services balance of the balance of payments, transactions with financial instruments, and the capital account.
Although the general linear trend of the balance of payments of Azerbaijan for 2012-2018 is characterized by a decrease, the increase in all items abovemen-tioned of the balance of payments from 2017 can be seen in fig.1.
The increase in revenues from the oil and gas sector has in the past again begun to ensure an increase in the total balance of payments surplus in Azerbaijan. In the second half of 2018, the trade balance surplus compared to the appropriate period in 2017 reached to $ 3.665 mlrd. with an increase of $ 1.462 billion, thus a
Рис. 1. Динамика ключевых показателей платежного баланса
Азербайджана в 2012—2018 годы (млн долл. США) Рис. 1. Динамика основных показателей платежного баланса Азербайджана в 2012—2018 гг. (млн долл. США)
Рис. 2. Приток и отток денежных средств в страну в 2017 и 2018 гг. по текущим операциям (млн долл. США)
Fig. 2. Cash inflows and outflows to the country in 2017 and 2018 on current transactions (million US dollars)
Таблица 1 (Table 1)
Динамика изменения торгового баланса в 2016—2019 гг. по сравнению с предыдущим соответствующим периодом (тыс. долл. США и в %)
Dynamics of trade balance changes in 2016-2019 compared to the previous relevant period (thousand US dollars and in%)
2016 2017 2018 2019
Total %-with Total %-with Total %-with Total %-with
Export 13210511 84,8 15152059 114,7 20793769 137,2 19868261 95,5
Import 9004176 92,1 9037316 100,4 10952441 121,2 11335316 103,5
Trade balans 4206335 -17,9 6114743 31,2 9841328 37,8 8532945 -15,3
Commersial turnover 22214687 - 16055795 -27,7 31746210 49,4 31203577 -1,73
Источник: Подготовлено автором на основе информации, полученной от Центрального банка Азербайджана (ЦБА)
Примечание: Знак «+» в табл. 1 указывает на увеличение, а знак «-» на уменьшение.
Source: Prepared by the author on the basis of information obtained from the Central Bank of Azerbaijan (CBA)
Note: The "+" sign in Table 1 indicates an increase, and the "-" sign indicates a decrease.
66% increase was achieved. During the period under review, revenues to the country reached to $ 21.407 billion, and outflows to $ 16.319 billion (see fig. 2). All statistics used in the analysis and graphs were taken from the official website of the Central Bank of Azerbaijan and State Statistical Committee of the Republic of Azerbaijan[14,15].
In 2018, compared to 2017, while the country's cash inflows from current operations increased in the amount of $ 5.416 billion or 33.9%, the increase in outflows from the country was $ 1.391 billion or 9.3%. As a result, cash inflows into the country are significantly higher than the outflows, the current account surplus has increased by almost 5 times and risen from $ 1.063 billion to $ 5.088 billion. According to experts, this factor played a decisive role in ensuring the stability of the country's national currency during the period under review.
Till the end of 2018, the trade turnover formed as a sum of import and export indicators reached a high level and increased to 31746 billion US dollars with an increase of 9.4%(see tab.1).
The main results of the study
In our initial research on modeling the dynamics of the balance of payments [16, 17], a regression analysis was conducted in order to conduct econometric analysis of the dependence of the current account of the balance of payments on total and foreign investments, exports and imports, the exchange rate of the Azerbaijani manat. Current account of y-balance of payments involved in the analysis, x1-foreign investment (FORINV), x2-export (EXP), x3-import (IMP), x4-ex-change rate of manat to US dollar (MANAT), x5-total investment (CENINV) are dependent and explanatory variables covering the years of 1995-2017.
In the study, it is important to pay attention to the Darbin Watson coefficients with increased
Таблица 2 (Table 2)
Результаты регрессионного анализа за 1995-2017 гг. (с пятью объясняющими факторами)
Results of regression analysis for 1995-2017 (with five explanatory factors)
FORINV EXP IMP MANAT CENINV C
Coefficient -1.041597 0.855962 0.040745 0.127631 -0.074593 -978.5116
Std. Error 0.214275 0.071390 0.054360 0.220403 0.144352 1126.953
t-Statistic -4.861035 11.98994 0.749535 0.579079 -0.516739 -0.868281
Prob 0.0001 0.0000 0.4638 0.5701 0.6120 0.3973
R-squared 0.986318 Mean dependent var 4332.196
Adjusted R-squared 0.982293 S.D. dependent var 6943.397
S.E. of regression 923.9354 Akaike info criterion 16.71462
Sum squared resid 14512163 Schwarz criterion 17.01084
Log likelihood -186.2181 Hannan-Quinn criter. 16.78912
F-statistic 245.0926 Durbin-Watson stat 2.928609
Prob(F-statistic) 0.000000
Таблица 3 (Table 3)
Результаты регрессионного анализа за 2006-2017 гг. (с пятью объясняющими факторами)
Results of regression analysis for 2006-2017(with five explanatory factors)
FORINV EXP IMP MANAT CENINV C
Coefficient -0.043786 0.940172 0.075568 -3669.255 -0.505909 -365.0545
Std. Error 0.851259 0.130751 0.097305 3126.497 0.394494 3458.028
t-Statistic -0.051436 7.190547 0.776614 -1.173600 -1.282425 -0.105567
Prob 0.9606 0.0004 0.4669 0.2850 0.2470
R-squared 0.977251 Mean dependent var 9106.875
Adjusted R-squared 0.958293 S.D. dependent var 6616.478
S.E. of regression 1351.229 Akaike info criterion 17.56227
Sum squared resid 10954912 Schwarz criterion 17.80472
Log likelihood -99.37361 Hannan-Quinn criter. 17.47250
F-statistic 51.54956 Durbin-Watson stat 2.922200
Prob(F-statistic) 0.000075
and very low results, expressed by a very high multidimensional coefficient of determination in time series. Thus, relying on a high coefficient of determination and ignoring the low Durbin Watson coefficient, which points to autocorrelation, can lead to "false" regression and express incorrect dependencies[18].
According to the results of regression analysis with the above parameters(see tab. 2), the number of observations was 23; R2 (determination coefficient) = 0.98; F-statistic (Fisher criterion) = 245.1; prob. = 0.00; DW (Durbin Watson statistics) = 2.92. The results obtained are quite satisfactory. The coefficient of determination indicates that the independent variables included in the model explain the dependent variable by 98%. Criterion F received a fairly reliable estimate with a high probability. However, the result obtained for the DW criterion can't be considered satisfactory. The critical limits for the DW criterion with n = 23 and k = 5 are DL = 0.90 and DU = 1.92. Since the calculated value of the DW criterion is greater than 2, the value 4 — DW = 1.08 is compared with critical values: DL < 1,08 < Dv. Alternatively, we get a analogical result: 4 - DU< DW< 4 - DL: because of being 2.08 < 2.92 <3.1, DW falls into the zone of uncertainty, and it is impossible to decide whether there is an autocorrelation.
The units of measurement of the independent variables xp x2, x3, x5 included in the model for regression analysis are expressed in US dollars. Taking into account the denomination of the Azerbaijani manat in 2006, the research period was shortened due to the fact that the exchange rate of the x4 manat, one of the explanatory factors was expressed in the national currency, which created a problem for the stability of the time series, the number of observations was reduced to 12 and covered the years of 20062017.
Number of observations according to the results (see Table 3) of the 2nd regression analysis 12; R2 (determination coefficient) = 0.97; F-statistic (Fisher criterion) = 51.5; prob = 0.000075; (Durbin Watson statistics) = 2.92. According to the results, no significant change is observed, so the quality of the model does not increase or decrease significantly, and DW statistics still fall into the zone of uncertainty. This doesn't tell us whether there is an autocorrelation over time. In such cases, steps such as extending the time series and editing the explanatory factors in the model can be used to improve the quality of the model.
In general, the current account deficit formed under the influence of the trade balance can be financed by capital inflows in the following forms: Foreign borrow-
ings from other countries, the International Monetary Fund, the World Bank; Assets sold to foreign investors; Direct investments that provide foreign exchange inflows into the country for the establishment of new production facilities; Foreign exchange reserves.
The application of these measures supports the reduction of foreign assets of the state. However, if the government increases its foreign debt, which significantly exceeds the current account deficit, then the country is in danger of a foreign debt crisis with the balance of payments. Proper regulation of these financial processes is very important for the balance of payments and the dynamic development of the country's economy in general.
Thus, in order to improve the quality of the model, we have ad-
justed it and included in the model an important macroeconomic financial indicator of the country, the factor of state budget deficit. The indicators we received from the Central Bank of Azerbaijan for the budget deficit cover the years of 2006-2017 and the unit of measurement is million manat: -85.5; 78.6; -79.6; -12.2; -178; -363.5; 306; -135; 352.8; -308.4; -286.5; -241.2. This indicator is included in the model as xg-budget deficit (BD). The changes in the regression model are shown in Table 4 below.
In the next step, the model includes a macroeconomic indicator of international foreign exchange reserves. Statistical data for official international foreign exchange reserves for 2006-2017 (in USD million) were obtained from the Central Bank and included in the model as follows: x^foreign exchange reserves(REZ; 1967.3; 4015.3; 6137; 5161.7; 6407.6; 10481.5; 11694.8; 14152; 13758.3; 5016.7; 3974.4; 5334.6.
The developmental dynamics of the explanatory factors in the model are described in a complex way in fig.3.
The model is formed as follows:
Y = -949.6 + 0,017FORINV + + 0,99EXP + 0,12IMP -- 3185,8MANAT -
- 0,55 CENINV + 2,64BD -- 0,12REZ
According to the results of the last regression analysis, the number of observations was 12; R2 = 0.98; F statistic = 30.95; prob. = 0.0025; DW = 2.66. The critical limits for the DW criterion with n = 12 and k = 7 are Dl = 0.103 and DU = 3.033. As the price of DW statistics is closer to 2 than the previous prices, the quality of the model is improving.
In order to build a successful model, it is important that the independent variables in the model, the regressors, have a sufficiently wide range of variations. The range of variation can be measured on the basis of variability
Таблица 4 (Table 4)
Результаты регрессионного анализа за 2006-2017 гг. (с шестью объясняющими факторами)
Results of regression analysis for 2006-2017(with six explanatory factors)
FORINV EXP IMP MANAT CENINV C
Coefficient 0.185987 0.995924 0.122918 -3744.041 -0.684332 -405.5565
Std. Error 0.862494 0.138169 0.104898 3074.143 0.420372 3399.490
t-Statistic 0.215638 7.208013 1.171791 -1.217914 -1.627923 -0.119299
Prob 0.8378 0.0008 0.2941 0.2776 0.1645 0.9097
R-squared 0.981681 Mean dependent var 9106.875
Adjusted R-squared 0.959698 S.D. dependent var 6616.478
S.E. of regression 1328.277 Akaike info criterion 17.51235
Sum squared resid 8821597. Schwarz criterion 17.79521
Log likelihood -98.07410 Hannan-Quinn criter. 17.40762
F-statistic 44.65689 Durbin-Watson stat 2.715850
Prob(F-statistic) 0.000348
Таблица 5 (Table 5)
Результаты регрессионного анализа за 2006-2017 гг. (с семью объясняющими факторами)
Results of regression analysis for 2006-2017 (with seven explanatory factors)
FORINV EXP IMP MANAT CENINV C
Coefficient 0.017248 0.997093 0.122752 -3185.879 -0.554208 2.640400
Std. Error 1.263255 0.153776 0.116669 4368.352 0.787641 2.618642
t-Statistic 0.013653 6.484073 1.052139 -0.729309 -0.703630 1.008309
Prob 0.9898 0.0029 0.3521 0.5062 0.5205 0.3703
R-squared 0.981872 Mean dependent var 9106.875
Adjusted R-squared 0.950148 S.D. dependent var 6616.478
S.E. of regression 1477.297 Akaike info criterion 17.66854
Sum squared resid 8729625 Schwarz criterion 17.99181
Log likelihood -98.01122 Hannan-Quinn criter. 17.54885
F-statistic 30.95050 Durbin-Watson stat 2.660883
Prob(F-statistic) 0.002511
Рис. 3. Динамика развития экономических параметров, входящих в модель на 2006-2017 гг.
Fig. 3. Development dynamics of economic parameters included in the model
for 2006-2017
(coefficient of variation), defined as the specific gravity of the ratio of standard variations of parameters to mathematical expectations:
vx = -x-100%. ' xi
a .
x V n -1
The results of the descriptive statistics in tab.6 were used to analyze the overall statistics and variability of the model. All re-
sults obtained should not be less than 10% to ensure variability, otherwise if this condition isn't met for any variable, it may be more appropriate to remove the variable from the model or replace it with another variable.
There is no need to replace all variables as the required condition is met.
A multicollinear analysis which expressing the correlation dependences between the explained and explanatory coefficients was also performed based on the Pearson coefficients. The calculated double-line correlation coefficients in tab.8 demonstrate strongly straight dependence between y and x5 (r = 0,74) and y and x2 (r = 0,92), strongly inverse dependence to y and x4 (r = -0,74), dependence at straight medium density between y and x7 (r = 0,55), weakly straight dependence between y and x6 (r = 0,32), and very weak inverse dependence between y and x1 (r = -0,11) and x3 (r = -0,08). All coefficients were calculated without taking into account the lags.
To evaluate the cause-and-effect relationships between the variables in our model, we performed the Granger test based on the Fisher criterion. According to the test, the H0 hypothesis rejects the existence of causal dependencies between all possible pairs in the model with Prob. (F-statistic) assumption[19].
Among the many results we obtained, the number of observations, F-Statistic and Prob(F-sta-tistic) is indicated in the tab.9 which can be a one-sided reason for the current account of the balance of payments and formulate based on the corresponding results that reject the H0 hypothesis. According to the F-Statistic criterion for the current account of the balance of payments, the pairs that confirm the H1 hypothesis with 95% probability are y- x1 (foreign investments), y- x2 (exports), y- x5 (total investments), y- x7 (foreign exchange reserves). Given the small number of obser-
Таблица 6 (Table 6)
Результаты описательной статистики Results of descriptive statistics
y x7 X6 x5 X4 x3 x2 x1
Mean 9106.875 7341.767 -79.37500 18121.17 0.977250 5201.833 23623.33 8626.508
Median 10298.20 5735.800 -110.2500 16906.70 0.812000 9020.500 23872.50 8897.200
Maximum 17146.10 14152.00 352.8000 27907.50 1.720000 10417.00 34494.00 11697.70
Minimum -1363.400 1967.300 -363.5000 8300.400 0.784000 -7574.000 13014.00 5052.800
Std. Dev. 6616.478 4091.665 229.6012 6393.992 0.329809 7062.213 8045.514 2190.651
Skewness -0.383181 0.595558 0.729514 0.279494 1.579096 -1.071375 -0.086722 -0.330311
Kurtosis 1.710498 1.931446 2.479581 1.891123 3.802741 2.306700 1.456093 1.819013
Jarque-Bera 1.125064 1.280283 1.199800 0.771038 5.309284 2.536019 1.206866 0.915576
Probability 0.569765 0.527218 0.548866 0.680098 0.070324 0.281391 0.546931 0.632682
Sum 109282.5 88101.20 -952.5000 217454.0 11.72700 62422.00 283480.0 103518.1
Sum Sq. Dev. 4.82E+08 1.84E+08 579883.9 4.50E+08 1.196514 5.49E+08 7.12E+08 52788467
Observations 12 12 12 12 12 12 12 12
Таблица 7 (Table 7) Коэффициенты вариации объясняющих переменных (в %) Coefficients of variation on explanatory variables (in%)
x x1 x2 x3 X4 x5 X6 x7
25,4 34 135,7 33,7 35,2 289,2 55,7
Таблица 8 (Table 8)
Корреляционная матрица Correlation matrix
y GENINV FORINV EXP IMP MANAT BD REZ
x5 x1 x2 x3 x4 x5 x7
y 1 0.74 -0.11 0.92 -0.08 -0.77 0.32 0.55
GENIN x5 0.74 1 0.07 0.76 0.26 -0.81 0.34 0.62
FORINV xj -0.11 0.07 1 0.23 0.74 0.21 0.09 0.65
EXP x2 0.92 0.76 0.23 1 0.16 -0.68 0.32 0.79
IMP x3 -0.08 0.26 0.74 0.16 1 0.22 -0.12 0.54
MANAT x4 -0.77 -0.81 0.21 -0.68 0.22 1 -0.46 -0.42
BD x6 0.32 0.34 0.09 0.32 -0.12 -0.46 1 0.4
REZ x7 0.55 0.62 0.65 0.79 0.54 -0.42 0.4 1
Таблица 9 (Table 9) Результаты тестов причинности Грейнджера (лаги:1) Results of Granger Causality Tests (lags: 1)
Null Hypothesis: Obs F-Statistic Prob.
*1 FORINV does not Granger Cause y 11 8.37980 0.0201
x2 EXP does not Granger Cause y 11 7.20144 0.0278
x, IMP does not Granger Cause y 11 1.09125 0.3267
x4 MANAT does not Granger Cause y 11 0.03827 0.8498
x, CENINV does not Granger Cause y 11 9.34623 0.0156
x6 BD does not Granger Cause y 11 0.48063 0.5078
^7 REZ does not Granger Cause y 11 6.41873 0.0351
vations (2006-2017, annual), the test doesn't allow for more lags, so we were able to conduct the test with only 1 lag. This necessitates an increase in the number of observations. Due to the current
technical problems with the provision of statistical information, we plan to increase the number of our observations in order to address this problem in our future research. For Y-X3(import), y- x4
Таблица 10 (Table 10) Расширенный тест Дики-Фуллера (с различиями 1-й степени) Extended Dickey-Fuller test (with 1st degree differences)
t-Statistic -3.107422 Log likelihood -85.97255
1% level -5.521860 F-statistic 6.662377
5% level -4.107833 Prob(F-statistic) 0.033771
10% level -3.515047 Mean dependent var -487.3889
Prob.* 0.1646 S.D. dependent var 8078.168
R-squared 0.799897 Akaike info criterion 19.99390
Adjusted R-squared 0.679835 Schwarz criterion 20.08156
S.E. of regression 4570.878 Hannan-Quinn criter. 19.80474
Sum squared resid 1.04E+08 Durbin-Watson stat 1.736959
Таблица 11 (Table 11) Расширенный тест Дики-Фуллера (с различиями 2-й степени) Extended Dickey-Fuller test (with 2nd degree differences)
t-Statistic -3.550973 Log likelihood -78.52002
1% level -4.582648 F-statistic 17.29797
5% level -3.320969 Prob(F-statistic) 0.005666
10% level -2.801384 Mean dependent var 2238.012
Prob.* 0.0370 S.D. dependent var 13326.02
R-squared 0.873724 Akaike info criterion 20.38000
Adjusted R-squared 0.823214 Schwarz criterion 20.40979
S.E. of regression 5603.045 Hannan-Quinn criter. 20.17908
Sum squared resid 1.57E+08 Durbin-Watson stat 1.570826
(manat) and y- x6 (budget deficit), the test results are not very satisfactory, because the H1 hypothesis can be accepted with very little probability.
In general, if autocorrelation is determined at the time series levels in the research process, it is important to eliminate it in some way before applying the regression equation for the forecast. If there is a strong autocorrelation between the levels, then it is better to use their differences in the calculations instead of quantitative indicators of the series. Differences in the form of yt = yt —
— yt-i; xti = xt,i — xt-i,i; ... xt7 =
= xt7 — xt-i 7 [20] are applied instead of the y, xi, X2, x3, x4, x5, x6, x7 variables in the model. The autoregression model for first-order differences is applied in the form of the following equation:
yt = «0 + axyt_i + st.
When the parameters of this model are calculated by the method of the smallest squares, the model is formed in a form of yt = a0 + a1yt-1 which is suitable for forecasting. It should be noted that it is purposeful to apply the method of differences on the basis of preliminary data when DW statistics approach 0 or 4[i8].
The stableness of the time series was tested on the basis of an extended Dickey-Fuller (DF) test. In this case, the H0 hypothesis accepts the assumption that the time series under study has a single root. In the initial stage of the test(see tab.iO), the H0 hypothesis is accepted that the time series has a single root for the current account balance(y). The stationary nature of the time series is not confirmed, so the value of the t-Statistic criterion is obtained with a very small probability t = -3.i07 withp = 0.i646. This indicates that the result obtained was observed with large errors, and the hypothesis H0, which confirms the single root, is not allowed to be rejected. At the same time, the fact that the test result for t-Statistic at the i%, 5%, and i0% significance levels is
to the left of the critical values for t-Statistic brings the time series closer to stationary. t = -3.107 value is to the right of three critical values, -5.521860(1%); -4.107833(5%); -3.515047(10%); It should be noted that the extended DF test was performed on the basis of the autoregression model(AR) and Schwarz criteria with constant and trend, 1st degree differences. The maximum number of lags was taken as 3. Due to this, the number of observations decreased to 9.
By changing the test parameters, the test in the next stage was conducted based on the autoregression model(AR) with 2nd degree differences, stable, without trends and here the maximum number of lags is 2(see tab.11).
From the results obtained in tab. 11, it can be seen that the stationary nature of the time series for the current account of the balance of payments can be accepted according to the results obtained. The value of the t-Sta-tistic criterion is obtained with the probability t = -3.55 with p = 0.037. This indicates that the
result occurred with minimal error and allows the H0 hypothesis to be rejected. The result obtained for t-Statistic is to the left respectively of -3.320969; -2.801384 critical values at the 5% and 10% significance levels. The adequacy of the autoregression model is also quite satisfactory. R2 = 0,87 indicates that the overall quality of the model is high. The corrected determination coefficient is 82%, the value F-sta-tistic = 17.3 was obtained with a high Prob(F-statistic) = 0.005 assumption. DW = 1.57 and is close to 2.
The main results.
A multidimensional regression model has been established that allows economic and statistical analysis of the dynamics of the current account of the balance of payments;
The form and directions of functional dependence between dependent and independent variables were determined, variability was assessed, the results of mul-tivariate regression analysis were
analyzed according to econometric methodologies; quantitative characteristics of the mechanisms of influence of explanatory factors on the balance of payments were measured and interpreted;
Correlation dependencies for cause-and-effect dependencies in the model were investigated, a Granger test was performed, and factors that significantly explained the outcome with high probabilities were identified based on the Fisher criterion;
The stability of the model was measured based on the Dick-
ey-Fuller test. With the first and second degree differences, the stableness of the autoregression model was determined on the basis of the Student's criterion by changing the lag sizes.
The built-in autoregression model demonstrates sufficient adequacy, the time series for the dependent variable is stationary, and can be considered suitable for the forecast values of the current account of the balance of payments.
The results of the research case provide an opportunity to
identify real development trends in the balance of payments of Azerbaijan at the present stage and to determine its interdependence with other macroe-conomic variables based on the analysis of the dynamics of the balance of payments and develop specific recommendations for balance of payments. The model also allows analyzing and forecasting the dynamics of Azerbaijan's national currency exchange rate and foreign economic activity within the balance of payments.
Литература
1. Mendoza E., Uribe M. The Business Cycles of Balance-of-Payments Crises: A Revision of a Mundellian Framework. NBER Working Paper. 1999. № 7045.
2. Mundell R. A. The Monetary Dynamics of International Adjustment under Fixed and Flexible Exchange Rates // The Quarterly Journal of Economics. 1960. Т. 74. № 2. С. 227-257.
3. Thirlwall A. P., Hussain M. N. The Balance of Payments Constraint, Capital Flows and Growth Rate Differences between Developing Countries // Oxford Economic papers. 1982. Т. 34. № 3. С. 498-510.
4. Moreno-Brid J. C. On Capital Flows and the Balance-of-Payments Constrained Growth Model // Journal of Post Keynesian Economics. 1998. Т. 21. № 2. С. 283-298.
5. Kumhof M., Li S., Yan I. Balance of Payments Crises under Inflation Targeting // Journal of International Economics. 2007. Т. 72. № 1. С. 242-264.
6. Cespedes L., Chang R., Velasco A. Balance Sheets, Exchange Rate Regimes, and Credible Monetary Policy [Электрон. ресурс] // Harvard University and NBER. 2001. Режим доступа: http://citeseerx.ist.psu.edu/viewdoc/download?doi = 10.1.1.203.1042&rep=rep1&type=pdf.
7. Кравцов М.К., Пашкевич А.В., Бурды-ко Н.М., Гаспадарец О.И. Система эконометри-ческих моделей для анализа и краткосрочного прогнозирования основных макроэкономических показателей Республики Беларусь // Экономика и управление. 2007. № 3. С. 69-80.
8. Харемза В.В., Харин Ю.С., Макарова С.Б., Малюгин В.И., Гурин А.С., Раскина Ю.В. О моделировании Экономики России и Беларуси на основе эконометрической модели LAM-3 // Прикладная эконометрика. 2006. № 2. С. 124-139.
9. Chaldaeva L.A., Chinaeva T.I., Bogopolskiy A.S. Analysis of financial and economic indicators, characterizing the activities of organizations in the oil and gas industry // Statistics and Economics. 2020. № 17(1). С. 69-78. DOI: 10.21686/25003925-2020-1-69-78/.
10. Ovcharov A.O., Terekhov A.M. Econometric analysis of the use of biological assets in agricultural
organizations // Statistics and Economics. 2020. № 17(1). С. 79-87. DOI: 10.21686/2500-39252020-1-79-87.
11. Pilnik N. P., Shaikhutdinova M. F. Modeling of the Balance of Payments State in Russia // Financial Journal. 2017. № 5. С. 84-101.
12. Engle, R. F. and Yoo, B. S. Cointegrated Economic Time Series: An Overview with New Results in R. F. Engle and C. W. J. Granger, eds., Long-Run Economic Relationships. Oxford: Oxford University Press, 1991. С. 237-266.
13. Johansen S. and Juselius K. Identification of the Long-run and the Short-run Structure: An Application to the ISLM Model // Journal of Econometrics.1994. № 63. С. 7-36.
14. Official website of the Central Bank of Azerbaijan [Электрон. ресурс]. Режим доступа: https://www.cbar.az; https://www.cbar.az/page-41/ macroeconomic-indicators.
15. State Statistical Committee of the Republic of Azerbaijan [Электрон. ресурс]. Режим доступа: https://www.cbar.az/page-41/macroeconomic-indicators.
16. Orudzhev Elshar G., Ayyubova Natavan S. Empirical analysis of the factors affecting the balance of payments in Azerbaijan // Aktual'ni Problemy Ekonomiky. Kiev. 2016. Т. 181. С. 400-411.
17. Айюбова Н.С., Исазаде А.Э. Вопросы моделирования и анализа взаимосвязей валютного курса маната и детерминантов национальной экономики Азербайджана // Евразийский Союз Ученых (ЕСУ), Ежемесячный научный журнал. 2021. Т. 6. № 1(82). С. 4-11. Серия: Экономические науки. ISSN 2411-6467. DOI: 10.31618/ ESU.2413-9335.2021.6.82.
18. Dickey D.A., Fuller W.A. Distribution of Estimators for Autoregressive Time Series with a Unit Root // Journal of the American Statistical Association.1979. № 74. С. 427-431.
19. Granger C.W.J., Newbold P. Spurious Regressions in Econometrics // Journal of Econometrics.1974. Т. 2. С. 111-117.
20. Kantorovich Grigory. Time series analysis // Higher School of Economics Economic Journal. 2002. Т. 6. № 2. С. 498-523
References
1. Mendoza E., Uribe M. The Business Cycles of Balance-of-Payments Crises: A Revision of a Mundellian Framework. NBER Working Paper. 1999; 7045.
2. Mundell R. A. The Monetary Dynamics of International Adjustment under Fixed and Flexible Exchange Rates. The Quarterly Journal of Economics. 1960; 74; 2: 227-257.
3. Thirlwall A. P., Hussain M. N. The Balance of Payments Constraint, Capital Flows and Growth Rate Differences between Developing Countries. Oxford Economic papers. 1982; 34; 3: 498-510.
4. Moreno-Brid J. C. On Capital Flows and the Balance-of-Payments Constrained Growth Model. Journal of Post Keynesian Economics. 1998; 21; 2: 283-298.
5. Kumhof M., Li S., Yan I. Balance of Payments Crises under Inflation Targeting. Journal of International Economics. 2007; 72; 1: 242-264.
6. Cespedes L., Chang R., Velasco A. Balance Sheets, Exchange Rate Regimes, and Credible Monetary Policy [Internet]. Harvard University and NBER. 2001. Available from: http://citeseerx.ist. psu.edu/viewdoc/download?doi=10.1.1.203.1042&r ep=rep1&type=pdf.
7. Kravtsov M.K., Pashkevich A.V., Burdyko N.M., Gaspadarets O.I. System of econometric models for analysis and short-term forecasting of the main macroeconomic indicators of the Republic of Belarus. Ekonomika i upravleniye = Economics and Management. 2007; 3: 69-80. (In Russ.)
8. Kharemza V.V., Kharin YU.S., Makarova S.B., Malyugin V.I., Gurin A.S., Raskina YU.V. On modeling the Economy of Russia and Belarus based on the LAM-3 econometric model. Prikladnaya ekonometrika = Applied Econometrics. 2006; 2: 124-139. (In Russ.)
9. Chaldaeva L.A., Chinaeva T.I., Bogopolskiy A.S. Analysis of financial and economic indicators, characterizing the activities of organizations in the oil and gas industry. Statistics and Economics. 2020; 17(1):69-78. DOI: 10.21686/2500-3925-2020-1-6978/. (In Russ.)
10. Ovcharov A.O., Terekhov A.M. Econometric analysis of the use of biological assets in agricultural
Сведения об авторе
Натаван Солтан Айюбова
К.э.н, доцент кафедры Математической экономики
Бакинский Государственный Университет,
Баку, Азербайджан
Эл. почта: neyyubova@mail.ru
organizations. Statistics and Economics. 2020; 17(1): 79-87. DOI: 10.21686/2500-3925-2020-179-87. (In Russ.)
11. Pilnik N. P., Shaikhutdinova M. F. Modeling of the Balance of Payments State in Russia. Financial Journal. 2017; 5: 84-101.
12. Engle, R. F. and Yoo, B. S. Cointegrated Economic Time Series: An Overview with New Results in R. F. Engle and C. W. J. Granger, eds., Long-Run Economic Relationships. Oxford: Oxford University Press; 1991: 237-266.
13. Johansen S. and Juselius K. Identification of the Long-run and the Short-run Structure: An Application to the ISLM Model. Journal of Econometrics.1994; 63: 7-36.
14. Official website of the Central Bank of Azerbaijan [Internet]. Available from: https:// www.cbar.az;https://www.cbar.az/page-41/ macroeconomic-indicators.
15. State Statistical Committee of the Republic of Azerbaijan [Internet]. Available from: https:// www.cbar.az/page-41/macroeconomic-indicators.
16. Orudzhev Elshar G., Ayyubova Natavan S. Empirical analysis of the factors affecting the balance of payments in Azerbaijan. Aktual'ni Problemy Ekonomiky. Kiev. 2016; 181: 400-411.
17. Ayyubova N.S., Isazade A.E. Issues of modeling and analysis of interrelations between the exchange rate of the manat and the determinants of the national economy of Azerbaijan. Yevraziyskiy Soyuz Uchenykh (YESU), Yezhemesyachnyy nauchnyy zhurnal = Eurasian Union of Scientists (ESU), Monthly scientific journal. 2021; 6; 1(82): 4-11. Series: Economic sciences. DOI: 10.31618/ ESU.2413-9335.2021.6.82. (In Russ.)
18. Dickey D.A., Fuller W.A. Distribution of Estimators for Autoregressive Time Series with a Unit Root. Journal of the American Statistical Association. 1979; 74: 427-431.
19. Granger C.W.J., Newbold P. Spurious Regressions in Econometrics. Journal of Econometrics.1974; 2: 111-117.
20. Kantorovich Grigory. Time series analysis. Higher School of Economics Economic Journal. 2002; 6; 2: 498-523.
Information about the author
Natavan S. Ayyubova
Cand. Sci. (Economics), Associate Professor of the
Department of Mathematical Economics,
Baku State Universitety,
Baku, Azerbaijan
E-mail: neyyubova@mail.ru
