AD ALTA
JOURNAL OF INTERDISCIPLINARY RESEARCH
lgREER
-2.5
-2.5
-3.0**
-3.3***
№ obs.
34
34
33
33
Critical values.
1%
-3.6
-4.2
-3.6
-4.3
5%
-2.9
-3.5
-2.9
-3.5
10%
-2.6
-3.2
-2.6
-3.2
PP test (H0: Unit root)
lgMI
-1.9
-3.2
-7.9*
-7.7*
lgOILP
-2.1
-2.8
-4.3*
-4.2**
lgOILQ
-4.5*
-6.8*
-16.0*
-16.4*
lgGE
-1.9
-1.8
-5.8*
-5.8*
lgCAS
-2.9
-4.5*
-8.5*
-8.3*
lgREER
-2.7
-2.7
-6.8*
-7.2*
№ obs.
37
37
36
36
Critical values.
1%
-3.6
-4.2
-3.6
-4.2
5%
-2.9
-3.5
-2.9
-3.5
10%
-2.6
-3.2
-2.6
-3.2
Notes: The symbols *, ** and *** refer to the rejection of the null hypothesis at the 1%, 5% and 10% significance levels,
correspondingly.
In our analysis, in order to capture the long run relationship
between the oil and gas sector and manufacturing industry of
Russia, VECM which is developed by Johansen (1988, 1995) is
employed (Johansen,1988; Johansen,1995). The model used in
the analysis has the following form:
∆
=Г
1
∆
−1
+Г
2
∆
−2
…+Г
−1
∆
−−1
+
′
−1
+
where ∆ is the difference operator, Z
t
is an (n × 1) vector of
variables= (MI
t
OILP
t
OILQ
t
GE
t
CAS
t
REER
t
D), k is the
number of lags, e
t
is an (n × 1) the vector of error terms. Г
refers to an (n × n) matrix of parameters, providing the
information about the short run relationship between variables. α
and β' are (n × r) adjustment and cointegration matrices,
respectively, which contains information regarding the long-run
relationships of series. Cointegration trace test and Max-Eig test
based on Johansen’s method are applied to verify the existence
of the cointegration and determine the number of cointegration
equations in the estimated model.
Finally, the Lagrange Multiplier test, proposed by Breusch and
Godfrey (1981), which permits to check the estimated model for
accuracy, is adopted (Breusch et al, 1981 ).
3 Results
The outcomes of the unit root tests which proved the integration
of order one I (1) of each series, permits us to conduct the
investigation for cointegration using Johansen’s methodology in
a multivariate framework. To capture the effect which is not
explained by the dependent variables, constant is included in
cointegration equation. Optimal lag length of 3 was determined
by using the sequential likelihood-ratio (LR) test, Akaike’s
information criterion (AIC) method and Schwarz Bayesian
information criterion (SBIC) method. The results of the Trace
test and Max-Eig test presented in Table 3 confirm the presence
of cointegration, namely our model cannot reject the null
hypothesis of at least one equilibrium cointegrating relation at
the 5% significance level. Based upon the results, it can be
concluded that the stable long run relationship between the
manufacturing industry (MI) and its explanatory variables exists
and it is time to proceed to the model estimation.
Table 3. Johansen cointegration test
Panel A. Trace test.
No. of cointegrating equations, r.
Trace statistic
5% critical value.
H
0
H
1
r = 0
r = 1
135.0*
104.9
r = 1
r = 2
76.9
77.7
r = 2
r = 3
40.2
54.6
Panel B. Maximum eigenvalue test.
No. of cointegrating equations, r.
Max-eigenvalue
statistic
5% critical value.
H
0
H
1
r = 0
r = 1
58.1*
42.4
r = 1
r = 2
35.6
36.4
r = 2
r = 3
17.5
30.3
Notes: The symbol * denotes rejection of null hypothesis at 5% significance level.
Table 4 presents the outcomes of the VECM estimation, where
the coefficient of manufacturing industry (MI) is normalized to
1. The parameters of explanatory variables are all statistically
significant and their values are reasonable with expected signs.
Generally, the cointegrating vector implies that MI depends
negatively on the OILP, OILQ, GE and REER, while CAS has a
positive effect on the development of manufacturing industry.
More specifically, 1% growth in oil price results in about 0.41%
decline of the manufacturing production, while the elasticity of
oil exports is considerably higher, namely 1% increase in the
volume of export decreases manufacturing production by more
than 2%. These findings support our expectations and can be
explained by the following postulates. Firstly, rise in oil price
and volume of crude oil export strengthens the real effective
exchange rate, which in turn negatively influences the
competitiveness of the Russia’s manufacturing products in the
international market, and as a result the total production goes
down. Another intuition that stays behind the obtained outcomes
is that growth in both variables (OILP and OILQ) has negative
impact on the domestic prices of raw materials used in
manufacturing industry. Subsequently, the price of finished
manufacturing goods goes up which further declines the demand
(1)
- 215 -