Purpose of research. Taking into consideration the abovementioned, an
econometric study of the dependence of Azerbaijan’s GDP from certain elements of
the economic inclusive parameters of the Russian Federation and the Republic of
Belarus in the time period from 1993 to 2018 is conducted. According to statistical
indicators, the dependence of Azerbaijan’s GDP from the balance of the trade relations
between Russia and Belarus is econometrically analyzed.
The linear multi-factor regression equality describing the dependency of
Azerbaijan’s GDP in thousands of USD (AZ_GDP _DOL) from the indicators of
thousands of USD balances of Russian Federation (RUSSIA_SALDO) and Belarus
(BELARUS_SALDO) will be as follows formally.
AZ_GDP_DOL = 29.6002446013*BELARUS_SALDO –
43.1759404878*RUSSIA_SALDO + 10733601.0684 (1)
The statistical information here were taken from the official electronic sources of
the State Statistics Committee of Azerbaijan [14].
Graph 1. Recursive residual estimates Graph 2. Standardized residual estimates
Checking the stability of the estimated parameters of the model over time is
carried out through the CUSUM test of Eviews package. Without taking into
account the residual, test results of residual limit models are shown properly Graph
-15
-10
-5
0
5
10
15
96
98
00
02
04
06
08
10
12
14
16
18
CUSUM
5% Significance
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
94
96
98
00
02
04
06
08
10
12
14
16
18
Standardized Residuals
SCIENTIFIC COLLECTION «INTERCONF» | № 3(39)
178
1 with standardized residuals on Graph 2. If on the Graph 1 the blue line is between red
lines, the Ho Hypothesis on stability of the parameters is accepted. If the blue line crosses
the red line, the Ho Hypothesis is declined and the alternative Ho Hypothesis on the
instability of parameters is accepted. The systematic variation of the regression equation
coefficients using the CUSUM Test is as follows (2). The graphs shown by examining
the stability of regression equation values with the CUSUM test in the studied range
show that the model residuals do not deviate beyond the 95% of confidence interval,
this means that the model's coefficients are consistent with the variation in the length of
the sample and it al-lows for high predictability of the model, making it possible to draw
conclusions about the stability of the model.
And now, let’s see the issue of presence of heteroscedasticity in the model. The
heteroscedasticity will lead to the ineffectiveness of the found assessment, as the
assessment will not be effective, although it will be coherent. We should mention that
in case of presence of heteroscedasticity, the values of standard errors found through
the least squares method (LSM) decrease, which result in the decrease in the value of
t-statistics and may result in inaccurate study about the significance of the assessment.
The heteroscedasticity may also occur due to the inaccurate selection of a model and
observation values. In case of presence of heteroscedasticity, the Summarized Least
Squares Method (SLSM) is applied [13] .
The heteroskedasticity was tested through the White test, and with 95%
probability, the hypothesis of homoskedasticity of the residues is not rejected. White’s
Test used second-order polynomials and pairwise products of factors.
The value
𝑛𝑅
2
=
𝑂𝑏𝑠 ∗ 𝑅 − 𝑠𝑞𝑢𝑎𝑟𝑒𝑑 , ,where 𝑛 = 26, 𝑅
2
- coefficient determination for auxiliary
regression of squared residuals on all regressors, their squares, pairwise products and
a constant is equal to
7.91, and this value with and this value is less than the
𝜒
0,16
2
(5) = 7.94.
Corresponding P-value
exceeds 0.05 the hypothesis of the
homoscedasticity of the random term is not rejected.
The model’s heteroscedasticity was checked through the White Test in the studied
model according to the observation results and the following results were achieved:
SCIENCE AND PRACTICE: IMPLEMENTATION TO MODERN SOCIETY
179
Do'stlaringiz bilan baham: |